How to Read a Credit Agreement

Introduction

A credit agreement is the most consequential document in leveraged finance and the one most juniors never actually open. It runs 250 to 500 pages, it is written in a dialect of English designed to survive litigation rather than to be read, and it governs everything: how much a company can borrow, what it can buy, what it can pay its owners, when it must hand cash back to lenders, and what happens when it stops performing. Every model you build of a leveraged company is a simplified restatement of that document.

The good news is that credit agreements are highly standardized. Once you have read three, you have effectively read the skeleton of every deal in the market, and the work becomes finding the handful of places where this deal departs from the template. The bad news is that those departures are exactly where the money is, and they are almost never in the sections a beginner reads first. Analysts open the document at the financial covenant, decide the deal looks tight or loose, and close it. Associates who know better open at Article I, because the definition of EBITDA can turn a 5.0x leverage covenant into an 8.0x leverage covenant without a single number in the covenant changing.

This walkthrough follows the order a junior banker actually reads the document on a live deal, not the order it is printed. Real agreements are public: they are filed as exhibits to 8-K and 10-K filings and are searchable through the SEC's EDGAR full-text search, which is the single best free training resource for anyone working in leveraged finance.

Almost every negotiated provision in a credit agreement is a compromise between two consistent sets of incentives. Lenders want visibility, early intervention rights, and cash coming back to them. Sponsors want operating freedom, capacity to do acquisitions and dividends, and as few consent requirements as possible. If you know which side of that trade a provision sits on, you can usually work out what it says before you read it.

Keep this table in your head as you read. When a provision looks unusually generous, the question is not whether it is generous but what the lender got in exchange for it elsewhere in the document.

The Standard Article Structure

Nearly all US credit agreements follow the same article order, which is why bankers can navigate an unfamiliar document in minutes. Article I holds definitions and accounting terms. Article II sets out the facilities themselves: commitments, borrowing mechanics, interest, fees, amortization, and mandatory prepayments. Article III contains representations and warranties, and Article IV lists conditions precedent to closing and to each subsequent borrowing.

The commercial heart begins at Article V, the affirmative covenants, and Article VI, the negative covenants. Article VII is events of default, Article VIII covers the administrative agent, Article IX is the miscellaneous article that quietly contains assignment rules and the amendment section, and the schedules and exhibits at the back carry the existing debt, existing liens, and disclosed subsidiaries. Some agreements insert financial covenants as their own article or bury them at the end of Article VI, which is why the table of contents is always the first page you read.

The Order to Read It In

Reading front to back is the slowest possible approach because Article I is meaningless until you know which terms matter. Experienced readers make one structural pass, then one commercial pass, then chase definitions only where the commercial pass flagged something.

Track questions in a running list as you go rather than stopping to resolve each one. Most of them resolve themselves two sections later, and the handful that do not are precisely what you bring to your VP.

Why EBITDA Is a Negotiated Number

Consolidated EBITDA in a credit agreement is not an accounting concept. It is a contractual construct, often running two to four pages, and every ratio in the document divides by it. Leverage covenants, incurrence tests, grower baskets, the pricing grid, and the excess cash flow calculation all sit downstream of that one definition, which is why a sponsor will trade almost anything to widen it.

The mechanics are simple. Start with net income, add back interest, taxes, depreciation, and amortization, then add a long list of negotiated items: restructuring charges, transaction expenses, integration costs, stock-based compensation, sponsor management fees, non-recurring losses, and projected cost savings. A company reporting $100 million of reported EBITDA can easily present $135 million of Consolidated EBITDA under its own credit agreement. At $500 million of debt, that is the difference between 5.0x and 3.7x leverage, and it is entirely a drafting outcome.

Consolidated EBITDA

Consolidated EBITDA is the contractually defined earnings measure used to test every ratio in a credit agreement. It starts from net income and adds back interest, taxes, depreciation, amortization, and a negotiated list of one-time charges and projected cost savings, so it is usually higher than the EBITDA a company reports to investors.

Understanding how bankers strip out genuine one-time items is a different exercise from reading a credit agreement definition, and the gap between them is worth knowing. Our walkthrough of how to normalize EBITDA for valuation purposes covers the analytical version; the credit agreement version is a negotiation, not an analysis.

Add-Back Caps and Run-Rate Synergies

The most contested add-back is projected cost savings, usually described as run-rate synergies. This lets the borrower count savings it expects to achieve but has not yet achieved, which converts a forecast into current-period earnings for covenant purposes. Market practice caps these adjustments in two ways: a percentage cap, commonly 20% to 25% of Consolidated EBITDA, and a realization window, typically 18 to 24 months from the action that generates the savings.

The details of that cap are where junior bankers earn their keep. Read for four things:

Uncapped run-rate add-backs are the single most aggressive feature in modern loan documentation. S&P Global Ratings has spent years arguing that projected savings systematically overstate the earnings companies actually deliver, a point it set out at length in its analysis of the EBITDA add-back fallacy.

The Definitions Behind the Definitions

EBITDA gets the attention, but three other defined terms decide almost as much. Indebtedness determines what counts as debt in the leverage ratio, and carve-outs for receivables facilities, sale-leasebacks, earnouts, or preferred equity can move leverage by a full turn. Restricted Subsidiary determines which entities are inside the credit group, and therefore whose EBITDA counts and whose assets secure the loans. Consolidated Total Assets is the reference point for many grower baskets, so a broad definition quietly expands capacity everywhere.

Watch for the phrase "greater of". A basket sized at the greater of $50 million and 25% of Consolidated EBITDA is a grower basket, and it only ever ratchets up. Watch too for "at the borrower's election" and "as determined by the borrower in good faith", both of which move discretion from the agent to management. Two definitions that appear identical across deals frequently differ by one clause, and that clause is usually the point of the negotiation.

Term Loan A Versus Term Loan B

Article II tells you what was actually lent. A term loan A is bank-held, typically five years, and amortizes meaningfully, often 5% to 10% of principal per year. It usually sits in the same credit facility as the revolver, shares its financial covenants, and is priced tighter because banks hold it to maturity and expect the relationship business that comes with it.

A term loan B is institutional. It is bought by CLOs, loan funds, and separately managed accounts, runs six to seven years, amortizes at a nominal 1% per year with a bullet at maturity, and is almost always covenant-lite. Pricing spans roughly SOFR plus 300 to 500 basis points depending on rating and market conditions, with single-B credits toward the wider end. Spreads compressed to multi-decade lows through 2025 before widening modestly in early 2026, so check current levels rather than carrying a number in your head. Deals typically come with an original issue discount and soft call protection of 101 for six months against repricing. When you read the tranche table, note the maturity gaps: a revolver that matures inside the term loan creates a springing maturity risk that shows up in the definitions rather than the facility description.

Revolvers, Delayed Draw, and Incremental Facilities

The revolver is the working capital and liquidity line. Read for the commitment size, the letter of credit sublimit, the swingline sublimit, the commitment fee on undrawn amounts, and whether the springing financial covenant tests at 30%, 35%, or 40% utilization. A springing covenant that only tests at high utilization is worth much less to lenders than it looks, because a borrower in trouble can simply stay below the trigger. If you are modeling the facility, our walkthrough of how to build a revolver and debt schedule with a cash sweep shows how these mechanics translate into the model.

Delayed draw term loans give the borrower committed capacity to fund a specific acquisition program within a set availability period, usually 12 to 24 months, and carry a ticking fee that steps up over time. Incremental facilities, often called the accordion, are the more important provision. They let the borrower add debt after closing without a new syndication, subject to a capacity formula that typically has three layers: a fixed "freebie" amount, a grower component tied to EBITDA, and an unlimited ratio-based amount available so long as pro forma leverage stays under a stated level.

MFN protection deserves particular attention. A 50 basis point MFN with an 18-month sunset and carve-outs for acquisition financing is materially weaker than a permanent MFN with no exceptions, and the difference determines whether existing lenders can be repriced out of the money by a later tranche.

Credit documents come up constantly in leveraged finance and restructuring interviews: Work through leveraged finance, LBO, and debt technicals with worked answers, start practicing interview questions for free and find the gaps before an interviewer does.

Reading the Margin Grid

Interest is set as a spread over SOFR, and in most bank-style facilities that spread moves on a pricing grid tied to leverage. A grid might charge SOFR plus 275 basis points above 4.0x, plus 250 between 3.5x and 4.0x, and plus 225 below 3.5x. The grid rewards deleveraging and, less obviously, penalizes the borrower exactly when it can least afford it, since leverage rises when EBITDA falls.

Three details matter more than the headline spread. First, the leverage definition used in the grid may be total, first lien, or net of cash, and net definitions with uncapped cash netting are meaningfully more borrower-friendly. Second, there is usually a floor on the base rate, commonly 0.00% to 0.75%, which matters when rates fall. Third, grid step-downs often do not apply for the first six months after closing, and a default typically freezes the borrower at the highest level. Institutional term loan Bs frequently have no grid at all, or a single step-down, because CLO investors prefer stable coupons.

The Excess Cash Flow Sweep and Other Mandatory Prepayments

Mandatory prepayments are the provisions that force cash back to lenders regardless of what management would prefer to do with it. The excess cash flow sweep is the one you will be asked about. It calculates cash generated after debt service, capital expenditure, taxes, and permitted investments, then requires the borrower to prepay a percentage of it, applied against the term loans.

Excess Cash Flow Sweep

An excess cash flow sweep is a credit agreement provision requiring a borrower to use a defined percentage of its surplus annual cash flow to repay term loan principal. The percentage typically starts at 50% or 75% and steps down to 25% or 0% as leverage falls below agreed thresholds, so a deleveraging company keeps more of its own cash.

Recent filed agreements show the pattern clearly: a 75% sweep stepping down to 50% when first lien net leverage falls below 4.00x and to 25% or zero at lower thresholds. What makes the provision softer than it sounds is the deduction list. Voluntary prepayments, permitted acquisitions, capital expenditure, and sometimes capital expenditure the borrower merely intends to make in the following year all reduce the sweep amount, and in strong markets sponsors negotiate the sweep down to a number that rarely bites.

The other mandatory prepayments follow the same logic. Asset sale proceeds above a threshold must be applied to the loans, but the borrower usually gets a reinvestment right of 12 to 18 months, extendable if it commits to reinvest within that window. Insurance and condemnation proceeds work the same way. Debt incurrence prepayments require 100% of proceeds from non-permitted debt, which in practice means the covenant baskets, not this clause, control new borrowing.

Affirmative Covenants and the Reporting Calendar

Article V is the least glamorous and the most operationally important part of the agreement for a borrower's finance team. It requires audited annual financials within 90 to 120 days, unaudited quarterlies within 45 to 60 days, an annual budget, and a compliance certificate showing the covenant calculations. It also requires maintenance of insurance, payment of taxes, preservation of corporate existence, and notice of defaults and material litigation.

Two affirmative covenants have real teeth. The first is the further assurances and additional guarantor covenant, which requires new material subsidiaries to join the guarantee and pledge their assets within a set number of days. The second is the going concern qualification clause, which turns an auditor's qualification into a default in many agreements. Both are frequently overlooked and both have triggered real defaults.

Negative Covenants and the Baskets That Matter

Article VI is where the document does its actual work. It prohibits categories of action and then permits them again through long lists of exceptions called baskets. Five covenants drive nearly every leveraged finance discussion:

The distinction between maintenance and incurrence testing shapes how these operate, and our explainer on maintenance versus incurrence covenants covers that split in detail. Inside each covenant, capacity comes in three forms: fixed baskets in hard dollars, grower baskets expressed as the greater of a dollar amount and a percentage of EBITDA or total assets, and ratio baskets that are unlimited so long as a leverage test is met on a pro forma basis.

Builder Basket

A builder basket, also called the available amount or cumulative credit, is a pool of restricted payment and investment capacity that accumulates over the life of a loan. It typically builds from 50% of cumulative consolidated net income or from retained excess cash flow, plus equity contributions, so a profitable borrower can pay dividends that the original basket sizes never contemplated.

Builder baskets are the reason a company can pay a large sponsor dividend years after closing without breaching anything. Grower baskets do something similar with less visibility: because they are sized off EBITDA, and EBITDA is itself a negotiated definition, aggressive add-backs quietly expand every basket in the document at the same time. When you are asked how much a borrower could dividend out, the honest answer is that you have to add the fixed basket, the grower basket, the builder basket, and the ratio basket together, then check whether they can be used cumulatively.

Financial Covenants, Headroom, and Equity Cures

Financial covenants set the ratio the borrower must maintain, and the number itself tells you less than the headroom behind it. Standard practice is to set the covenant with 25% to 35% cushion to the sponsor's base case model, so a company projecting 4.0x leverage at close gets a 5.5x covenant. Step-downs then tighten the level over time, usually in quarterly or semi-annual increments, on the assumption that the business deleverages.

Equity cure rights let the sponsor inject cash that counts as EBITDA (or, in stronger documents, only as debt reduction) to fix a breach, typically limited to two cures in any four quarters and five over the life of the loan. In broadly syndicated loans the financial covenant is often only a springing test on the revolver, which is why the Federal Reserve has repeatedly flagged the vulnerability of firms leaning on floating-rate leveraged loans and private credit in its May 2026 Financial Stability Report.

Article VII lists what constitutes a default and what lenders can do about it. The categories are consistent: payment default on principal (usually no grace period) or interest (typically one to five business days), breach of a financial covenant, breach of a negative covenant (often no cure period), breach of an affirmative covenant with a 30-day cure after notice, material misrepresentation, bankruptcy, unsatisfied judgments above a threshold, ERISA events, change of control, and invalidity of the guarantees or liens.

Cross-default is the provision worth understanding properly. A cross-default clause triggers a default under this agreement when the borrower defaults on other debt above a threshold, even if that other lender takes no action. A cross-acceleration clause is narrower and only triggers when the other lender actually accelerates. Sponsors push hard for cross-acceleration with a high dollar threshold, because it means a technical breach on a small equipment lease cannot cascade into a default on the entire capital structure.

Remedies also matter more than analysts expect. Most defaults are not automatic: the required lenders must vote to accelerate, and in practice they rarely do, because acceleration crystallizes losses. What actually happens is a waiver or amendment negotiation with a fee, a spread increase, and tighter terms, which is the outcome the default provisions are really designed to produce.

Who Guarantees and What Is Excluded

A leveraged loan is only as good as the assets and cash flows that stand behind it. The security documents grant a first priority lien on substantially all assets of the borrower and each guarantor, and the guarantee covenant requires all material wholly owned domestic restricted subsidiaries to guarantee. The interesting part is always the exclusion list: immaterial subsidiaries below a revenue or asset threshold, foreign subsidiaries, joint ventures, regulated entities, captive insurance vehicles, and non-wholly owned subsidiaries.

Guarantor coverage is usually expressed as a percentage of consolidated EBITDA and assets, commonly 80% to 90%. Read for what sits in the other 10% to 20%, because that is where value can migrate. Foreign subsidiary exclusions are especially worth checking in businesses where international operations are growing faster than the domestic base, since the guarantor coverage test can be satisfied at closing and quietly deteriorate afterward.

Unrestricted Subsidiaries and the Intercreditor

The single most dangerous concept in the modern credit agreement is the ability to designate a subsidiary as unrestricted. An unrestricted subsidiary is outside the credit group entirely: it is not bound by the covenants, its assets are not collateral, and its EBITDA does not count toward the ratios.

Unrestricted Subsidiary

An unrestricted subsidiary is an entity that a borrower designates as sitting outside its credit group. It is not subject to the credit agreement's covenants, its assets do not secure the loans, and it can incur its own debt, which is why designation capacity and the investment baskets that fund it are among the most heavily negotiated provisions in leveraged finance.

Designation consumes investment basket capacity, so the practical question is how much capacity exists and whether intellectual property or other crown jewel assets can be transferred out. The intercreditor agreement, usually a separate document referenced in the credit agreement, then governs the relationship between first lien and second lien or unsecured creditors: payment waterfalls, standstill periods (typically 90 to 180 days before junior creditors can enforce), voting on plan support, and rights in bankruptcy. It is short relative to the credit agreement and disproportionately important when things go wrong.

Debt documents are one of the highest-yield technical topics in leveraged finance recruiting: Our 160-page PDF walks through leveraged finance, LBO mechanics, and the questions interviewers ask, access the IB Interview Guide to prepare properly.

Article IX contains the voting mechanics, and it is the last section most people read and the first section any restructuring lawyer opens. Ordinary amendments require Required Lenders, defined as holders of more than 50% of outstanding loans and commitments, which means a majority can change most of the document. A narrower set of provisions, known as sacred rights, requires the consent of each affected lender: reductions in principal or interest, extensions of maturity, releases of all or substantially all collateral or guarantees, and changes to the pro rata sharing and waterfall provisions.

Read for who counts in the vote. Sponsor affiliates and debt funds connected to the sponsor are usually disenfranchised, but the size of the permitted affiliate holding and whether it votes on amendments is negotiated. Also read the assignment provisions in the same article: disqualified lender lists, borrower consent rights over transfers, and open market purchase permissions all determine who can accumulate a blocking position in a distressed situation.

How the Loopholes Became Liability Management

Everything above stopped being academic in 2016. J.Crew used its investment baskets to transfer intellectual property to an unrestricted subsidiary and borrowed against it, leaving existing lenders without the collateral they thought they had. Serta Simmons used the "open market purchase" exception in its assignment provisions to do a deal with a majority group of lenders that lifted them to a new senior tranche, subordinating everyone else. Both transactions were built entirely from provisions that had been sitting in standard documents for years.

The market response has been documentary. Blockers named after the deals that inspired them are now common: J.Crew blockers restrict transfers of material intellectual property to unrestricted subsidiaries, and Serta blockers make lien subordination a sacred right. Surveys of loan documentation show the sacred right requiring affected lender consent to subordinate moved sharply: before 2020 only about 40% of loans restricted uptiers at all, often inadvertently, while by mid-2022 85% explicitly blocked them and 70% required unanimous lender consent to subordinate. In Europe, 89% of 2025 leveraged loans for buyouts and full refinancings carried some form of J.Crew protection, against 44% of European high yield bonds, though presence is not quality: on one review only about a tenth of those loan blockers were drafted tightly enough to work as intended. The Fifth Circuit's December 2024 decision that Serta's uptier was not an open market purchase pushed the market further in the same direction, though drafting keeps evolving around each new precedent. Our breakdown of uptier and drop-down liability management transactions covers the mechanics in full.

Reading a credit agreement well is mostly an exercise in knowing what is worth escalating. Nobody wants a 12-page summary. They want the five things that are different from market and what those differences cost.

Present each item the same way: what the provision says, what it permits in dollars or turns of leverage, and what the market standard is. That framing is what separates an analyst who read the document from one who understood it. It is also, almost verbatim, how these questions get asked in leveraged finance and restructuring interviews.

Credit agreements reward repetition more than talent. The tenth one takes an hour, and the difference between an analyst who is genuinely useful on a financing and one who is not usually comes down to whether they have done that work.

Start with a filed agreement for a company you already understand, read it in the order above, and write down every question you cannot answer. Do that three times and the fourth document will look like a variation on a theme you already know, which is exactly how the people negotiating them see it.