Debt Service Coverage Ratio (DSCR) in 2026: How Lenders Calculate It, Where It Breaks, and What the Covenant Really Says

Quick answer: The debt service coverage ratio (DSCR) measures how many times a borrower's cash flow covers its scheduled debt payments: DSCR = cash flow available for debt service ÷ (interest + scheduled principal). A DSCR of 1.25x means $1.25 of cash flow for every $1.00 of debt service. Commercial lenders typically require 1.10x–1.25x at origination, and in credit agreements DSCR is tested quarterly as a covenant whose result depends on how the agreement defines each input.

The ratio is easy to compute. Agreeing on what goes into it is the hard part.

Search for "DSCR" and most results come from residential mortgage lenders: monthly rent divided by the monthly mortgage payment, a minimum of 1.0x or 1.25x, and a calculator. That's a real product, the investor DSCR loan, but it isn't what a credit analyst means when a borrower's compliance certificate shows a debt service coverage ratio of 1.31x against a 1.20x covenant.

In commercial, private credit and project-finance lending, DSCR is a negotiated covenant whose value is produced by its definition. The same borrower, with the same accounts, can pass under one agreement's definition and fail under another. This guide covers the ratio from the lender's side: the variants, the contract language, current thresholds, breach data, and the specific mistakes that inflate it.

Why DSCR means something different to a credit analyst

The mortgage version of DSCR is a qualification screen: gross rent ÷ PITIA (principal, interest, taxes, insurance and association dues), with floors usually between 1.00x and 1.25x. Some consumer lenders advertise programs that accept ratios below 1.0x [17]. It's tested once, at origination, by the lender.

The commercial version differs on three counts:

1. It's defined in a contract. The credit agreement specifies what counts as cash flow, what counts as debt service, which add-backs are allowed and capped, and how accounting changes are handled.
2. It's tested repeatedly. Maintenance covenants are typically tested every quarter, on a trailing-twelve-month basis, from a stated date.
3. The borrower calculates it. The compliance certificate is prepared and signed by the borrower's financial officer. The lender's job is to check it.

That third point is where the work, and the risk, sits. The OCC's commercial real estate handbook states that covenant-compliance DSCR may differ from the DSCR used for underwriting and risk rating, and that income and expense definitions should be spelled out in the loan documents [1]. Put simply, regulators expect at least two DSCRs to exist for the same loan.

The DSCR formula and its variants

Every version shares one skeleton:

DSCR = cash flow available for debt service ÷ scheduled debt service (interest + principal)

Below 1.0x, cash flow doesn't cover debt service. At 1.0x, it covers it exactly. Above 1.0x, there's headroom. The variants differ in what goes on each side of the line.

What a DSCR of 1.25x actually means

A DSCR is easiest to judge as a cushion: how far cash flow can fall before it no longer covers debt service. The cushion equals 1 − (1 ÷ DSCR).

So the common 1.25x requirement means a borrower can absorb a 20% drop in cash flow, assuming debt service stays fixed, before payments are no longer covered. For floating-rate debt, rising rates shrink the cushion from the other side.

Sources: OCC [1]; SBA SOP 50 10 [3]; Energy Resources 12 credit agreement [6]; CFADS modelling guidance [19].

NOI vs EBITDA

CRE DSCR deducts a replacement reserve whether or not it's funded. EBITDA-based DSCR adds back depreciation, which is the economic cost the reserve is meant to capture, and doesn't deduct capex unless the definition says so. The same property can therefore look stronger on an EBITDA test than on an NOI test. The OCC uses the NOI basis, with the reserve deducted, as its supervisory baseline [1].

EBITDA vs CFADS

In project finance the gap is large enough to change a credit decision. In one worked toll-road year, EBITDA is 33.6 and CFADS is 28.0 after cash taxes, working capital and maintenance spending, against debt service of 21.5. CFADS gives a DSCR of 1.30x; dividing EBITDA directly reports 1.56x, overstating coverage by about a fifth [19].

Entity-level vs global

It isn't enough for the business to cover the new loan. The OCC states that cash flow should cover all of the borrower's obligations, not just the loan being underwritten [1], and the SBA requires 1:1 global coverage alongside the entity-level test [3].

A worked example: one borrower, three answers

The table below uses illustrative figures for a single borrower over the same trailing twelve months. Only the definition changes.

Nothing about the business changed between rows. That's why the first question about any reported DSCR is which definition produced it.

How to size a loan from a required DSCR

Lenders often run the formula in reverse to find the maximum loan a property or business can support:

1. Maximum annual debt service = cash flow ÷ required DSCR
2. Maximum loan = maximum annual debt service ÷ mortgage constant (the annual payment per dollar of loan, which depends on rate and amortization)

With $600,000 of NOI and a 1.25x requirement, maximum debt service is $480,000. At a mortgage constant of about 8.1%, that supports a loan of roughly $5.9 million [23]. Raise the requirement to 1.35x and the same NOI supports about $5.5 million, which is why a 0.10x change in the threshold matters so much in practice.

What DSCR looks like inside a real credit agreement

None of the top search results for "DSCR" quote an actual covenant. Here are two filed in 2024 on SEC EDGAR.

A true DSCR maintenance covenant: Energy Resources 12 (2024)

This oil and gas credit agreement prohibits the Debt Service Coverage Ratio from being "less than 1.20 to 1.0" at the end of any fiscal quarter, starting with the quarter ending June 30, 2024 [6]. Three features matter more than the 1.20x:

• The numerator is EBITDAX less distributions. Deducting distributions inside the ratio works as a built-in lock-up: the more the sponsor takes out, the closer the borrower gets to breach.
• Principal isn't taken from the amortization schedule. It's the annualized amount needed to amortize the loan over the half-life of the borrower's oil and gas properties, plus other current maturities.
• The accounting is frozen. Covenant calculations ignore ASC 842 lease reclassification, and must be made consistently with prior periods unless the parties agree otherwise in writing.

Compliance is borrower-certified: each financial officer's certificate must show reasonably detailed calculations, and breach of the financial covenants is an immediate Event of Default with no grace period [6].

A bank fixed charge coverage test: Link Media / First National Bank of Omaha (February 2024)

This Eighth Amendment defines the Consolidated Fixed Charge Coverage Ratio as EBITDA minus unfinanced maintenance capex, cash income taxes and dividends, divided by cash interest, scheduled principal and capital-lease payments [7]. Two specific dividends are carved out of the numerator by name and date. The facility had been amended eight times between 2019 and 2024, which shows how often coverage definitions get re-cut in practice.

Definitions that recurrently cause disputes

Disagreements rarely come from the ratio itself. They come from the defined terms it points to:

• "Indebtedness." One agreement lists twelve categories, including capital leases, letters of credit, guarantees, hedge termination values and receivables factoring; others exclude swap obligations or carve out standby letters of credit up to a dollar cap. Every carve-out is a place where debt can drop out of the denominator.
• "Interest." Some definitions include amortized fees, original issue discount and the interest element of capital-lease rent; others are strictly cash-basis.
• "Fixed Charges." At least one filed definition of Fixed Charges covers interest and preferred dividends only, with no scheduled principal, so a ratio labelled "fixed charge coverage" can be less demanding than a DSCR.
• Pro forma and test-period mechanics. Acquisitions after the measurement period may be excluded from the maintenance covenant but included for incurrence tests, and some agreements lock the test date for a pending transaction.
• Add-back caps. Caps exist because uncapped adjustments are material; one 2021 agreement capped restructuring costs, litigation expense and cost savings at 20% of pre-add-back EBITDA [8]. For the full picture on adjustments, see our guide to EBITDA add-backs in 2026.

Minimum DSCR thresholds by loan type in 2026

There's no single "good" DSCR. The answer depends on the loan type, the jurisdiction, and whether you mean an origination floor, a covenant level or a rating threshold.

SBA: the only segment with published program floors

The SBA's lending rules changed on October 1, 2026, when SOP 50 10 8.1 replaced SOP 50 10 8 [3][4].

The acquisition change is the one that matters most commercially. Under the new rules, projections may be reviewed but can't carry an initial acquisition, and one annotated comparison estimates that moving from 1.15x to 1.25x reduces supportable debt by roughly 8% [5]. The SBA also now requires lenders to review two months of bank activity to confirm every business debt is in the denominator, and a lender-engaged quality of earnings report for larger acquisitions and expansions [5]. Both requirements turn DSCR into a document-heavy exercise; see how AI-assisted commercial finance underwriting handles bank statements, debt schedules and financials in one package.

Program-level figures above draw on a clause-referenced annotated comparison of the two SOPs [5]. Check the SBA's SOP text before relying on a specific threshold.

Commercial real estate and multifamily

The OCC doesn't set a single number. It requires the ratio to reflect amortization and cash-flow volatility: lower ratios can be prudent for stable long-term net leases, while hotels and owner-occupied properties with uneven earnings warrant higher ones [1]. Market practice clusters as follows:

Sources: industry glossaries and program terms [18][22]. Property-type levels other than the Freddie Mac program come from secondary sources, not supervisory text.

Two details separate institutional underwriting from consumer pages. Interest-only periods don't improve coverage for a bank: the OCC requires interest-only loans to meet debt service requirements as if they amortized [1]. And lender NOI is usually lower than borrower NOI, because the lender applies market vacancy, management fees, leasing costs and a replacement reserve even when actual results are better.

Because loan size is calculated backwards from the required DSCR (see how to size a loan from a required DSCR), these property-type differences translate directly into how much can be borrowed against the same income.

Bank C&I, leveraged lending and private credit

There's no regulator-set DSCR for commercial and industrial loans. The OCC's leveraged lending guidance expects minimum interest coverage and fixed charge coverage covenants alongside maximum leverage [2]. Filed agreements show fixed charge coverage tests between 1.00x and 1.25x, often stepping up over time or springing only when revolver availability drops.

Private credit doesn't publish floors, but the realized numbers are telling. In Lincoln International's US portfolio, size-weighted fixed charge coverage averaged 1.23x in Q2 2025, yet 25.5% of companies were below 1.0x, while covenant defaults ran at only 3.4% [11]. If a quarter of borrowers sit below 1.0x but only a few percent are in covenant default, then covenant thresholds are set on far looser definitions than the reported metric, or many deals have no maintenance coverage test at all. Covenant-lite structures are increasingly common in private credit for stronger borrowers [21]. Where coverage covenants are weak or absent, lender protection shifts to other parts of the document; our guide to liability management exercises covers what happens when those protections are tested.

Project finance

Project-finance DSCR is CFADS divided by scheduled debt service, measured for every period to maturity. S&P generally requires a forecast minimum DSCR of at least 1.05x for a "BB" category project [9]. Lenders typically size debt to a higher target and treat around 1.10x as the real minimum below which the asset is in trouble [20]. A contracted 50 MWp solar case at a constant 1.37x is described as comfortably within the market range [24]. Distribution tests trap cash when DSCR falls below the lock-up level, and debt service reserve accounts are commonly sized at about six months of debt service.

India, the UK and the EU

India is the main market with regulator-quoted numbers: the RBI's 2020 resolution framework set a DSCR floor of 1.00 and an average DSCR of 1.20 where no sector-specific threshold applies, with sector exceptions [14]. The RBI's 2025 project-finance directions impose no DSCR minimum, but require lenders to record DSCR in a project-finance database and update it within 15 days of any change [15].

The UK and EU have no regulator-set numeric DSCR floor in the evidence we reviewed. The practical difference is documentation: UK and European mid-market terms have converged with the syndicated market, including wider acceptance of covenant-lite structures [21].

How often coverage covenants are breached

Default indices measure different things, which is why the same quarter can produce very different numbers.

Sources: Proskauer [12]; Lincoln International [11][13]; Morgan Stanley [16].

Proskauer counts bankruptcies, payment defaults, distressed exchanges and covenant defaults together and weights by loan count. Lincoln counts covenant defaults only and weights by size. Both remove a breach once it's resolved [16]. Neither is wrong, but they shouldn't be quoted side by side without saying so.

What happens after a breach is more consistent. Academic research finds that about 15% of non-investment-grade or unrated borrowers breach a covenant each year. When a resolution event occurs, about 36% of firms get a pre-emptive amendment, 23% a waiver, and 21% an amendment after the breach; technical default is comparatively rare. Where a DSCR covenant is involved, the threshold usually moves down: from 1.64x to 1.53x on average after an amendment with a violation, and from 1.61x to 1.55x after a waiver [10].

Coverage covenants are rarely enforced to acceleration; they're renegotiated. The covenant's practical value is as an early warning and a seat at the table, which makes the accuracy of each quarter's calculation all the more important.

Why two lenders get different DSCRs: common calculation mistakes

Most DSCR errors are mechanical and can be quantified.

Sources: CFADS modelling guidance [19]; SBA SOP 50 10 8.1 [5]; OCC [1]; SEC filings [6][7].

A non-obvious one in project finance: with no maintenance reserve, a project can show 1.37x in a normal year and 0.81x in a major-maintenance year. The average looks healthy; the bad year is a payment default unless a reserve covers it [19].

Most of these errors show up as inconsistencies across documents (a debt in the bank statements that's missing from the schedule, a lease in the notes that's missing from fixed charges), which is what a risks and red flags review across the full borrower file is designed to catch.

Borrower-reported vs lender-recomputed DSCR

The compliance certificate is the borrower's calculation, so it carries the borrower's definitional choices and the borrower's errors. Some agreements acknowledge this directly: if a certificate is later found to be inaccurate and the corrected figure would have produced a higher interest rate, the borrower must deliver corrected statements and pay the difference.

Independent recomputation is standard practice elsewhere in finance. In third-party due diligence on mortgage pools, S&P treats DSCR as a critical data field with a 10 basis point materiality threshold, and may require diligence on the full pool if material errors reach 5% of the sample [9]. A credit team monitoring a private credit portfolio rarely holds itself to that standard on every certificate, but recomputing coverage from the underlying financials, rather than copying the borrower's number into a tracker, is the closest equivalent.

DSCR vs FCCR vs interest coverage ratio

The three coverage ratios answer different questions.

Lenders choose based on where repayment comes from:

• DSCR when the cash flow belongs to a ring-fenced asset and debt is sized against it: a property, a project, an SBA-financed business.
• FCCR when the borrower is an operating company with large fixed costs outside debt service. A lease-heavy retailer can post a comfortable DSCR and a tight FCCR.
• ICR for pricing, incurrence capacity and comparability. It's the relevant test only where the facility doesn't amortize.

How common each one is matters too. In US public covenant packages, leverage covenants appear at about 72% of firms, FCCR at 44% and interest coverage at 42%; DSCR appears at about 8%, with a median threshold of 1.25x [10]. DSCR is the specialist tool of asset-based, project and government-guaranteed lending, not the default corporate covenant.

One more caution: FCCR isn't simply "DSCR plus rent." At least one filed definition of Fixed Charges excludes scheduled principal entirely, so the label tells you less than the definition does.

How to review a DSCR covenant, step by step

Use this as a first-pass checklist for each test date.

1. Extract the definition, not just the number. Capture the DSCR definition and every defined term it references (EBITDA, Indebtedness, Interest, Fixed Charges), plus caps, carve-outs, pro forma rules and any accounting freeze. Our free Financial Covenant Extractor pulls covenant levels and definitions from a credit agreement in one pass.
2. Check the amendment history. Coverage definitions are re-cut often; Link Media's facility was amended eight times in five years. Make sure you're applying the definition currently in force.
3. Rebuild the numerator from source. Start from the financial statements, apply only the add-backs the agreement permits, and enforce the caps. Our EBITDA Add-back Analyzer flags adjustments that exceed common caps.
4. Rebuild the denominator from a complete debt schedule. Reconcile against bank statements, lease schedules and guarantees; check that new debt is included and that excluded items match the definition. On a new deal, this belongs in your wider due diligence work.
5. Recompute, then reconcile to the certificate. Compare your number to the borrower's. Each difference should trace to a specific definitional choice or input.
6. Measure headroom and stress it. Calculate the cushion to the covenant level and test it against a revenue decline and a rate increase. In one project-finance example, a 10% revenue fall took DSCR from 1.30x to 1.13x [19].
7. Log the outcome. Record the result, headroom, any waivers or amendments, and the sources behind each input, so the next quarter starts from evidence rather than memory.

Where AI-assisted review fits

AI tools are now good at the first step and unreliable at the fifth. An academic covenant-extraction system reports 92.17% test accuracy on classifying covenants from SEC filings [10]. A March 2026 vendor benchmark of LLMs on credit-agreement due diligence reports 85.11% aggregate accuracy for the best platform-enhanced model, but individual debt tasks ranged from about 53% to 100% [25]. That benchmark was run by the vendor, which discloses the conflict. No source we found publishes an accuracy rate for computing a DSCR end to end from a full set of financials. (For how we test our own system, see Continua's benchmarks; for a market view, see our comparison of AI underwriting platforms.)

Independent research on LLMs in capital markets gives the clearest design rule: use the model to parse documents and resolve definitions, compute in deterministic code, and keep a human sign-off [26]. In short, AI can find the definition faster than an analyst, but it shouldn't decide whether the borrower is in breach.

That's the approach behind Continua's credit review workspace: every extracted definition and figure links back to the page and clause it came from, the complete agreement and its amendments are read rather than sampled, and conflicts between documents, such as a certificate figure that doesn't match the financials, are flagged for the analyst to resolve. To check a borrower's numbers against each other, see Find Contradictions.

FAQ

What is a good debt service coverage ratio?

It depends on the loan. For commercial real estate, 1.20x–1.25x is a common minimum, higher for hotels and specialty property. SBA loans under SOP 50 10 8.1 require 1.10x to 1.25x depending on the program. Project finance commonly targets 1.25x–1.40x with a minimum near 1.05x–1.10x. In covenants, what matters more than the level is the definition behind it.

How do you calculate DSCR?

Divide cash flow available for debt service by scheduled debt service (interest plus principal) for the same period. In a credit agreement, use the defined terms exactly: the permitted EBITDA add-backs and caps, the deductions for capex, taxes or distributions, and the debts and payments included in debt service.

What happens if a borrower's DSCR falls below the covenant?

A maintenance covenant breach is typically an Event of Default, sometimes immediately with no grace period. In practice most breaches end in a waiver or amendment rather than acceleration, often with a lower threshold, higher pricing, fees, an equity injection or tighter reporting. Some agreements also allow an equity cure.

What is the difference between DSCR and FCCR?

Both include interest and scheduled principal in the denominator. FCCR typically also includes rent, lease payments and other fixed charges, and deducts unfinanced capex, cash taxes and distributions from the numerator, so it's usually the stricter test for lease-heavy operating companies. Definitions vary, so check the agreement.

What is global DSCR?

Global DSCR combines the business's cash flow with the cash flow of owners, guarantors or affiliates, and divides it by all of their combined debt service. The SBA requires at least 1:1 global coverage alongside the entity-level test, and the OCC expects cash flow to cover all of the borrower's obligations, not just the new loan.

Does an interest-only period improve DSCR?

On a consumer DSCR mortgage, it can help a borrower qualify. In institutional underwriting it generally doesn't: the OCC requires interest-only loans to meet debt service requirements as if they amortized, and agency programs such as Freddie Mac's small-balance loans calculate coverage on an amortizing payment.

Conclusion

DSCR looks like the simplest ratio in credit: cash flow over debt service. In a commercial or private credit agreement, it's a negotiated definition, calculated by the borrower, tested quarterly, and usually renegotiated rather than enforced when it breaks. The useful questions aren't "what's a good DSCR?" but "which definition produced this number, what did it leave out, and does it match what we get when we rebuild it from the source documents?"

Lenders who recompute coverage from the agreement's definitions, rather than accepting the certificate figure, catch missing debt, stretched add-backs and mismatched periods before they turn into a surprise waiver request.

Check a definition now: paste a credit agreement's coverage covenant into the free Financial Covenant Extractor to pull the threshold and every defined term it depends on. No sign-up required.

Sources

1. OCC, Comptroller's Handbook: Commercial Real Estate Lending, Version 2.0, March 2022. Link
2. OCC, Comptroller's Handbook: Leveraged Lending, with March 2025 annotations. Link
3. SBA, SOP 50 10: Lender and Development Company Loan Programs (version list, including SOP 50 10 8.1 effective October 1, 2026). Link
4. SBA, Information Notice 5000-880695: Issuance of SOP 50 10 8.1, August 14, 2026. Link
5. PCA, SOP 50 10 8 vs SOP 50 10 8.1: The Complete Annotated Comparison (secondary, clause-referenced). Link
6. Energy Resources 12, L.P., Credit Agreement (2024), SEC EDGAR. Link
7. Link Media Holdings, LLC, Eighth Amendment to Credit Agreement with First National Bank of Omaha, February 15, 2024, SEC EDGAR. Link
8. PLBY Group / Playboy Enterprises, Credit Agreement, May 25, 2021, SEC EDGAR. Link
9. S&P Global Ratings, Project Finance Operations Phase criteria (republished October 11, 2021), and proposed US RMBS criteria (2024). Link · Link
10. CovenantAI working paper, Sydney Banking and Financial Stability Conference, 2025 (covenant prevalence, breach and reset data, and extraction accuracy). Link
11. Lincoln International, Private Market Perspectives: U.S. Edition, Q2 2025. Link
12. Proskauer, Private Credit Default Index, Q2 2025, Q3 2025 and Q4 2025 releases. Link · Link · Link
13. Lincoln International, European Lincoln Private Market Index, Q1 2026. Link
14. Reserve Bank of India, Resolution Framework for COVID-19-related Stress, September 7, 2020. Link
15. Reserve Bank of India, Project Finance Directions, 2025, effective October 1, 2025. Link
16. Morgan Stanley Investment Management, Private Credit Market Monitor, Q2 2026. Link
17. AD Mortgage, DSCR Loan program page. Link
18. Commercial Real Estate Loans, DSCR glossary. Link
19. Julian R. Sterling, How do you calculate CFADS in project finance? Link
20. Breaking Into Wall Street, DSCR and LLCR (project finance). Link
21. Latham & Watkins, Private Credit 2025. Link
22. CBRE, Freddie Mac Conventional Small program summary, April 2026. Link
23. Stack CRE, DSCR and Commercial Loan Sizing. Link
24. PKF Deutschland, finance metrics, issue 5/2026. Link
25. ToltIQ, A Comprehensive Analysis of Platform-Enhanced LLM Accuracy Across 20 Critical Due Diligence Use Cases, March 2026 (vendor study; conflict disclosed). Link
26. Capital Markets LLM Reliability Score (CM-LRS), arXiv 2607.21340, July 2026. Link

This article is for general information only and isn't legal, credit or investment advice. Covenant terms vary by agreement; always rely on the executed credit documents and current regulatory text.

Start a free investigation · ← Back to all articles