Regulatory figures cited reflect Consumer Financial Protection Bureau rules in effect as of 2026; market pricing figures are labeled with their data year inline and vary by lender, credit profile, and location. This article is educational and is not mortgage or legal advice.
TL;DR — Quick Verdict
- Page 2 of the Loan Estimate is where the money is. Section A (Origination Charges) and Section B (Services You Cannot Shop For) are locked at 0% tolerance under CFPB Regulation Z — a lender that quotes them low and raises them later must refund the difference at closing.
- Section C (Services You Can Shop For) carries no tolerance limit if you use your own provider, which is the single largest source of surprise increases between estimate and closing.
- The interest rate on page 1 is not binding unless the box marked “can change” is unchecked — an unlocked rate can move before closing with no disclosure violation.
- APR is the only figure that lets you compare two lenders honestly, and it is accurate only within 0.125% under Regulation Z §1026.22 for most fixed-rate loans.
- In a modeled comparison of two lenders quoting the same 6.75% rate, total five-year cost differed by roughly $4,800 because of points and origination structure alone.
- Request Loan Estimates from at least three lenders within a 14-day window, then compare Section A and the Total Interest Percentage — not the monthly payment.
Most borrowers spend under four minutes reading a document that governs a 30-year financial commitment. That imbalance is expensive. The Consumer Financial Protection Bureau built the three-page Loan Estimate specifically so that two competing offers could be laid side by side and compared line for line — yet the majority of applicants still compare only the monthly payment, which is the least informative number on the form.
This article walks through the nine figures on a Loan Estimate that determine what a mortgage actually costs: which boxes are legally locked, which can rise without limit, how to read the Comparisons section on page 3, and how to spot a lender quoting an artificially low estimate to win your business. We model two real-world offers from lenders such as Rocket Mortgage and a regional credit union to show how a 0.25% difference in points can outweigh a rate advantage. Every tolerance figure cited comes from Regulation Z as published by the CFPB, and every market range is labeled with its data year.
The Three Tolerance Buckets That Decide What You Actually Pay
Federal rules sort every fee on your Loan Estimate into one of three tolerance categories, and the category — not the dollar amount — determines whether a lender can raise it before closing. This is the structural fact almost no borrower knows, and it is printed nowhere on the form itself.
Under Regulation Z §1026.19(e)(3), a lender that exceeds an applicable tolerance must refund the excess to the borrower no later than 60 days after consummation. That refund obligation is what gives page 2 of the Loan Estimate its teeth.
Source: Consumer Financial Protection Bureau, Regulation Z §1026.19(e)(3), TILA-RESPA Integrated Disclosure Rule (verify at consumerfinance.gov).
The asymmetry is deliberate. Regulators locked down the fees a lender controls and left flexible the fees driven by third parties and calendar timing. A borrower who understands prepaid insurance, tax, and interest at closing will not be alarmed when Section F moves by several hundred dollars — that movement is expected and legal.
Page 1: The Four Numbers Above the Fold
Loan Amount, Interest Rate, Monthly Principal & Interest, and Prepayment Penalty sit at the top of page 1 in a grid with a critical second column headed “Can this amount increase after closing?” That column is the entire point of the grid, and it is routinely ignored.
Consider a scenario. A borrower in Ohio receives an estimate showing 6.625% on a $340,000 loan. The rate box says “NO” under can-this-increase — but a separate line at the top of the page reads “Rate lock: NO, unless you lock, your interest rate and closing costs are subject to change.” Those two statements are not contradictory. The “NO” means the rate cannot change after closing; the lock line means it can absolutely change before closing. Borrowers conflate these constantly, and a 0.375% drift during a slow underwriting review and closing delays adds roughly $80 per month on that loan balance.
Mortgage rates through 2025 and into 2026 have moved within a range that makes float-versus-lock a real decision rather than an academic one; Freddie Mac’s Primary Mortgage Market Survey publishes the weekly benchmark, and borrowers should pull the current week’s figure rather than rely on a rate quoted in a general article. Period-specific July 2026 survey data was unavailable at publication.
Prepayment penalty deserves one sentence of attention: on a conforming conventional loan it is almost always “NO,” and if it says “YES,” treat that as a reason to request estimates elsewhere before proceeding.
Section A vs Section B: Where Lenders Compete and Where They Cannot
Two boxes on page 2 both carry 0% tolerance, but they behave completely differently in a shopping context. Understanding the difference is what separates a borrower who saves $2,000 from one who does not.
Section A contains what the lender charges to make the loan: discount points, an origination fee components and negotiability, application and underwriting fees, and processing charges. Every dollar here is revenue the lender sets, which means every dollar here is negotiable — and a competing Loan Estimate is the strongest negotiating instrument available.
Section B contains third-party services the lender requires and selects: the appraisal, the credit report pull, flood zone determination, and sometimes a tax service fee. The lender does not profit from these but must estimate them accurately. Appraisal fees on single-family conventional loans generally fell in the $500–$800 range nationally in 2025, with metro and complex-property assignments running higher; national point-figure data specific to 2026 was not available from a primary source at publication. Our detail on home appraisal cost, process, and low appraisals covers the drivers behind that spread.
Verdict
Negotiate Section A aggressively and ignore Section B entirely. A lender who discounts Section A by $1,200 to match a competitor is giving up margin; a lender who “discounts” Section B is either absorbing a cost they will recover elsewhere or under-quoting a fee they are legally required to eat. When comparing two Loan Estimates, subtract Section B from both and compare only Section A plus points — that comparison is apples to apples, and it is where the real spread lives.
Comparing Two Real Offers: The $4,800 Spread Hidden Behind an Identical Rate
Rate shopping breaks down because two lenders can quote the same number and charge wildly different amounts to get there. Below is a modeled comparison on a $400,000, 30-year fixed loan, both offers quoting 6.75%.
Modeled illustration by Real Cost Report using standard amortization on a $400,000 30-year fixed loan. Payment figures calculated by the authors; fee structures reflect typical 2025 lender pricing patterns. Rate benchmark context: Freddie Mac Primary Mortgage Market Survey (verify at freddiemac.com).
The five-year spread is roughly $4,800, and it exists entirely because Lender A charged points to reach a rate Lender B offered without them. Both Loan Estimates are honest. Both show the same rate and the same payment. Only Section A reveals the difference.
Verdict
Lender B wins decisively for any borrower who will sell, refinance, or pay off within seven years — which describes most 30-year mortgage holders. Lender A’s points only break even if the loan survives past roughly year eight at the same rate. If you cannot state with confidence that you will hold this exact loan for a decade, take the no-points offer and put the $5,000 toward principal, reserves, or the down payment instead.
Page 3: Total Interest Percentage and the Comparisons Box
Buried at the bottom of the third page sits the most analytically useful figure the CFPB requires anyone to disclose: Total Interest Percentage, or TIP. It expresses total interest paid over the full loan term as a percentage of the original loan amount — a 6.75% 30-year loan carries a TIP near 133%, meaning the borrower pays $1.33 in interest for every dollar borrowed.
Next to it sits APR, which folds Section A charges and certain other finance charges into an effective annualized rate. Regulation Z §1026.22 requires APR accuracy within 0.125% for regular transactions, which is tight enough that a meaningful APR gap between two offers is a genuine cost gap, not a rounding artifact. Where two lenders quote the same nominal rate but differ by 0.15% in APR, the higher-APR lender is charging more in Section A — full stop.
The Comparisons box also shows “In 5 Years,” a figure combining total payments and principal paid down over 60 months. Use it. It captures the points-versus-rate trade-off automatically, and it is the fastest legitimate shortcut on the entire form.
One caution: APR does not include Section C shopping items, owner’s vs lender’s title insurance, or the initial deposit into your escrow account calculation and payment changes. Two loans with identical APRs can still differ by $1,500 in cash required at the table.
What Most People Get Wrong Reading a Loan Estimate
Five errors account for the majority of avoidable cost at closing, and each has a specific corrective action.
Mistake 1: Comparing monthly payments across lenders
Payments differ because escrow estimates differ, not because the loan is cheaper. One lender may estimate property taxes conservatively and another aggressively, producing a $140 monthly gap on identical loans. Correct action: compare Section A totals and the “In 5 Years” figure, never the payment.
Mistake 2: Treating the Loan Estimate as an approval
It is a pricing disclosure issued after six pieces of information, not an underwriting decision. Conditions surface weeks later, and mortgage denial causes and reapplication costs can run into thousands once appraisal and application fees are sunk. Correct action: treat the estimate as a quote and the conditional approval as the milestone.
Mistake 3: Shopping title and settlement independently without checking tolerance
Selecting a provider not on the lender’s written list moves that fee from the 10% aggregate bucket to no tolerance at all. Correct action: if you shop, get a written quote from your chosen provider before removing the tolerance protection.
Mistake 4: Ignoring the “Estimated Cash to Close” line
Borrowers focus on closing costs and forget that cash to close includes the down payment, prepaids, escrow funding, and any seller concession limits toward closing costs credited back. Correct action: reconcile that single number against your actual liquid funds before proceeding.
Mistake 5: Not requesting a revised estimate after a changed circumstance
Rate locks, appraisal outcomes, and loan program changes all permit a lender to reset tolerance baselines — but the lender must issue a revised Loan Estimate within three business days of the triggering event. Correct action: ask for the revised disclosure in writing every time something material changes.
Who Needs to Scrutinize This Document Hardest
Not every borrower faces the same exposure. The value of a line-by-line review scales with three variables: loan size, how many lenders you are actually willing to walk away from, and how unusual your property or income is.
Scrutinize hardest if you are a self-employed borrower or a retiree drawing from assets rather than wages. Both profiles generate more lender-side underwriting work, which shows up as higher Section A charges and longer closing timelines and what affects them. On a $600,000 loan, a 0.5-point difference in origination pricing is $3,000 — enough to justify several hours of comparison work.
Condo and planned-development buyers carry a second layer, since HOA financial review in underwriting can add fees and delays that never appear on the initial estimate at all.
A borrower purchasing at or below the county conforming limit with a W-2 income, 780 credit score, and 20% down has a genuinely commoditized loan. Three estimates and a Section A comparison will capture nearly all available savings in under an hour. Anyone outside that profile should expect wider dispersion between lenders and should collect four or five estimates. Reviewing which closing costs are negotiable before the first call materially improves the outcome of that conversation.
Frequently Asked Questions
How long is a Loan Estimate valid?
Under CFPB rules, the lender must honor the estimated closing costs for a minimum of 10 business days from the date the Loan Estimate is provided, giving you a defined shopping window. The interest rate is not covered by that 10-day window — rates remain subject to change until you formally lock. After 10 business days, the lender may issue a revised estimate with different figures without any changed circumstance.
Does requesting multiple Loan Estimates hurt my credit score?
Mortgage inquiries made within a focused shopping window are treated as a single inquiry by the major credit scoring models — FICO uses a 45-day window for its newer versions and 14 days for older ones still used in mortgage lending. Because many lenders pull older FICO versions, compressing all applications into 14 days is the safer approach. The score impact of one mortgage inquiry is typically under five points.
What happens if my closing costs exceed the tolerance limits?
Regulation Z §1026.19(f)(2)(v) requires the lender to refund the excess to you no later than 60 calendar days after consummation, along with a corrected Closing Disclosure. In practice most lenders apply the cure as a credit on the Closing Disclosure itself. If you spot a Section A or Section B increase at the closing table with no valid changed circumstance, raise it before signing — the correction is far easier pre-closing.
Can I get a Loan Estimate without a property address?
No. A lender is only required to issue a Loan Estimate once it receives all six elements of a complete application: name, income, Social Security number, property address, property value estimate, and requested loan amount. Before that, any figures you receive are an informal worksheet or pre-qualification quote carrying no tolerance protection whatsoever — a distinction worth confirming in writing with any lender presenting numbers early.
How We Researched This Article
Regulatory figures in this article were drawn directly from the text of Regulation Z as published and interpreted by the Consumer Financial Protection Bureau, specifically the provisions governing good-faith estimate tolerances, revised disclosure triggers, APR accuracy tolerances, and the Loan Estimate delivery timeline established by the TILA-RESPA Integrated Disclosure rule. Where the regulation itself specifies a numeric threshold — 0%, 10%, 0.125%, 60 days, three business days, 10 business days — that figure is quoted from the rule rather than from any secondary summary.
Interest rate context was checked against the Freddie Mac Primary Mortgage Market Survey, the longest-running weekly national rate benchmark. We did not publish a point rate figure for the current week, because rate levels change weekly and any figure printed here would mislead a reader arriving later; borrowers should pull the survey directly. Pricing-adjustment context reflects the loan-level price adjustment framework published by Fannie Mae for conventional conforming loans.
The two-lender comparison is modeled, not measured. Monthly principal and interest figures were calculated by the authors using standard amortization on a $400,000 30-year fixed loan at 6.75%; the fee structures represent common lender pricing patterns rather than quotes from any named institution. No borrower’s actual Loan Estimate was used. Readers should reproduce the calculation with their own loan amount and quoted rates rather than transferring our five-year totals to their situation.
Limitations are meaningful. Appraisal fee ranges and national closing cost averages cited reflect 2025 data because primary-source figures specific to 2026 were not available at publication; these vary substantially by state, property type, and metro area. Fee ranges describe conventional single-family transactions and do not extend to FHA, VA, USDA, jumbo, or investment-property loans, each of which carries a different cost structure. Additional context on typical fee composition was reviewed through U.S. Department of Housing and Urban Development settlement cost materials.
Research for this article was last conducted in July 2026. All figures were verified against named primary sources before publication.