The IRS Mileage Log Sampling Method (Three-Month Rule) for Freelancers 2026

Q: Does the IRS let you keep a mileage log for only part of the year?

Yes, in limited circumstances. Treasury Regulation §1.274-5(j)(2) allows the 'sampling method' — keeping adequate records for part of the year and using them to substantiate business use for the entire year — but only if you can establish that the sample period is representative of the whole year. The regulation's own example is a taxpayer who keeps a log for the first week of each month, or for a representative portion of the year, and reasonably extrapolates the rest. The key word is representative: if the sample doesn't reflect your typical driving, the method fails.

Q: What is the three-month mileage sample rule?

It's a common application of the IRS sampling method: instead of logging every trip for 12 months, you keep a complete, detailed mileage log for one representative three-month period (often a calendar quarter), calculate the business-use percentage for that period, and apply it to your full-year mileage from odometer readings. It works only if those three months genuinely reflect your normal pattern — a freelancer whose work is steady year-round can use it; a seasonal driver whose busy quarter isn't typical cannot.

Q: Do I still need odometer readings if I use the sampling method?

Yes. The sample gives you a business-use percentage, but you need total annual miles to turn that percentage into a deduction. Photograph or record your odometer at the start and end of the year (and ideally at the start and end of the sample period). Total annual miles is the denominator; without it, a percentage is meaningless and the extrapolation can't be computed or defended.

Q: Why can't seasonal or holiday-heavy freelancers use a three-month sample?

Because the sample must be representative of the entire year. A rideshare driver who works mostly during summer festivals, a tax preparer slammed January–April, or a wedding photographer busy only in peak season would produce a wildly inaccurate full-year number if they sampled their high or low season. For these freelancers the IRS expects a year-round log, or at minimum a sample structured to capture the seasonal mix (such as one representative week per month across all 12 months).

Q: Is the sampling method risky in an audit?

It carries more risk than a complete year-round log. Vehicle expenses fall under the heightened substantiation rules of IRC §274(d), which means the Cohan rule — estimating a deduction without records — does not apply. If an auditor decides your sample wasn't representative, the entire extrapolation can be thrown out and your vehicle deduction reduced or denied. The method is legitimate, but only when the sample is genuinely typical and you've documented why. A complete log is always the safer choice.

Q: Can a mileage app do the sampling method for me?

An app can make a full year-round log so easy that you may not need to sample at all — which is the safest outcome. If you do sample, an app helps by capturing the four required elements (date, miles, destination, purpose) accurately during the sample period and by storing odometer photos for the total-miles calculation. The extrapolation math is simple; the hard part is proving the sample is representative, which is a recordkeeping discipline no app can fully replace.

Published: May 27, 2026 · Reading time: 7 min

TL;DR: The IRS sampling method under Treas. Reg. §1.274-5(j)(2) lets you keep a mileage log for only part of the year — a representative three-month stretch, or the first week of each month — and extrapolate business use across the full year. It works only if the sample is representative of your normal driving. You still need year-start and year-end odometer readings to get total annual miles, and you must document why the sample is typical. Because vehicle expenses fall under §274(d) heightened substantiation, there's no Cohan-rule relief — if an auditor finds the sample unrepresentative, the whole extrapolation can be tossed. At the 2026 rate of $0.725/mile, the method can save real time, but a complete year-round log is always safer.

Nobody enjoys logging every business trip for 365 days. The good news: the IRS doesn't strictly require it. Buried in the regulations is the sampling method — substantiate part of the year, extrapolate the rest. The bad news: freelancers misuse it constantly, sampling their busiest or slowest months and producing numbers an auditor can dismantle in minutes. This guide explains exactly how the sampling method works, when you can use it, when you can't, and how to run a sample that holds up.

What the Sampling Method Actually Says

The authority is Treas. Reg. §1.274-5(j)(2), which states that a taxpayer who keeps adequate records for part of a tax year may use them to substantiate business use for the entire year — provided the records, supplemented by other evidence, are representative of the whole year.

The regulation gives two classic patterns:

The recurring-week sample — keep a complete log for the first week of each month all year. This naturally captures seasonality.
The block sample — keep a complete log for a representative portion of the year (commonly a three-month quarter) and extrapolate, if business use is consistent across the year.

Either way, the governing principle is representativeness, not duration. Three months is convenient, not magic.

The Two Things You Must Have

The sampling method produces a percentage, not a deduction. To turn that percentage into dollars you need both:

Ingredient	Where it comes from
Business-use %	Your detailed sample-period log
Total annual miles	Year-start and year-end odometer readings

Photograph your odometer on the first and last day of the year. Then:

Business miles = total annual miles × sample business-use % Deduction = business miles × $0.725 (2026 rate)

Without total annual miles, a percentage is just a number floating in space. See the track business mileage / IRS requirements guide for the four data points every trip entry needs.

A Worked Example

Priya freelances as a home-organizing consultant with a steady, year-round client load. She uses a three-month sample:

Sample period (Q1): 3,000 total miles, of which 1,800 were to client jobs and supply runs
Sample business-use %: 1,800 ÷ 3,000 = 60%
Total annual miles (odometer): 20,000
Extrapolated business miles: 60% × 20,000 = 12,000
2026 deduction: 12,000 × $0.725 = $8,700

Because her client volume is consistent month to month, the Q1 sample reasonably represents her whole year — and she keeps a note saying so. If audited, that note plus the odometer photos and the detailed Q1 log defend the $8,700.

When You CANNOT Use a Three-Month Sample

The method collapses the moment your sample isn't typical. Watch out if you're:

Seasonal — a wedding photographer, festival rideshare driver, or holiday-market seller whose busy quarter dwarfs the rest
Spiky — a tax preparer slammed January–April, or a consultant with a single huge multi-month project
Newly ramping — a freelancer whose business grew sharply mid-year, so early months look nothing like later ones

For these freelancers, a block sample of one quarter produces a wildly wrong full-year number. The fix is either a full year-round log or the first-week-of-each-month sample, which spreads across all 12 months and captures the seasonal mix. The commuting vs business miles guide also matters here — make sure the miles in your sample are actually deductible business miles, not commuting.

Why the Method Is Riskier Than a Full Log

Vehicle and mileage deductions sit under the heightened substantiation rules of IRC §274(d). That has a sharp consequence: the Cohan rule — which lets you estimate ordinary expenses without perfect records — does not apply to mileage. There is no "reasonable estimate" fallback. See the Cohan rule and lost receipts guide for what the rule does and doesn't cover.

So if an auditor concludes your sample wasn't representative:

The extrapolation is rejected
You're left with only the actual logged sample miles — a fraction of your claim
The rest of the deduction is denied, with interest and possibly penalties

That's the trade-off. The sampling method is real and legitimate, but it converts a recordkeeping shortcut into an audit argument about representativeness. A complete log never has that argument.

Sampling vs a Full Log vs an App

Approach	Effort	Audit strength
Full year-round log	Highest	Strongest — no representativeness question
First-week-of-each-month sample	Medium	Strong if seasonality is captured
Three-month block sample	Lower	OK only if business use is consistent
Reconstructed-from-memory log	Low	Weakest — often rejected under §274(d)

The honest takeaway: the sampling method earns its keep only if a full log is genuinely impractical for you. With a phone-based tracker, the effort gap between "sample" and "full log" has nearly vanished — which is why most freelancers are better off just logging everything. Compare the methods that feed your deduction in the standard mileage vs actual expense guide, and confirm the rate in the 2026 IRS mileage rate guide.

How CentSense Makes Sampling (or Skipping It) Easy

CentSense captures each business drive with the date, mileage, destination, and purpose §274(d) requires — and stores odometer photos alongside your receipts. That makes a clean three-month sample effortless to keep and makes a full year-round log so low-friction you may not need to sample at all. Either way, the extrapolation math and the supporting evidence live in one export at year-end.

Frequently Asked Questions

Does the IRS let you keep a mileage log for only part of the year?

Yes, in limited circumstances. Treas. Reg. §1.274-5(j)(2) allows the sampling method — keeping adequate records for part of the year to substantiate the whole year — but only if the sample is representative of the full year. The regulation's example is a log kept for the first week of each month, or a representative portion of the year, reasonably extrapolated. If the sample isn't representative, the method fails.

What is the three-month mileage sample rule?

It's a common application of the sampling method: keep a complete log for one representative three-month period, compute its business-use percentage, and apply it to your full-year odometer miles. It works only if those three months reflect your normal pattern — fine for steady year-round work, not for seasonal driving.

Do I still need odometer readings if I use the sampling method?

Yes. The sample gives a business-use percentage; you need total annual miles (from year-start and year-end odometer readings) to turn that percentage into a deduction. Without total miles, the percentage can't be applied or defended.

Why can't seasonal or holiday-heavy freelancers use a three-month sample?

Because the sample must represent the whole year. A driver busy only in peak season, or a preparer slammed January–April, would produce a wildly inaccurate full-year number from a block sample. They need a year-round log or a first-week-of-each-month sample that captures the seasonal mix.

Is the sampling method risky in an audit?

More than a full log. Vehicle expenses fall under IRC §274(d), so the Cohan rule (estimating without records) doesn't apply. If an auditor finds the sample unrepresentative, the entire extrapolation can be thrown out and the deduction denied. The method is legitimate only when the sample is genuinely typical and documented.

Can a mileage app do the sampling method for me?

An app can make a full year-round log so easy you may not need to sample — the safest outcome. If you do sample, it captures the four required elements during the sample period and stores odometer photos for the total-miles math. The hard part — proving the sample is representative — is a recordkeeping discipline no app fully replaces.

Authoritative References

Log Every Mile Without the Spreadsheet

CentSense captures each business drive — date, miles, destination, purpose — the moment it happens, and keeps the odometer photos and receipts that back it up. Whether you sample a quarter or log the whole year, your mileage deduction is audit-ready, not reconstructed in April. The Solo plan ($5/month) includes unlimited AI receipt scanning, mileage logging at the 2026 IRS rate of $0.725/mile, and a CPA-ready CSV export.

Start free →

This guide is general education for U.S. freelancers and Schedule C filers in 2026. It is not personalized tax advice — bring your specific facts to a CPA or EA for a complete return.

Compare alternatives

See how CentSense stacks up to other expense and receipt tools for freelancers.