Lost Your Mileage Log? How to Rebuild One for the CRA
How to reconstruct a mileage log from records made at the time, recover the odometer, map distances, and what the Tax Court has accepted and rejected.
Most people find out they needed a vehicle logbook at one of two moments: while filling in the motor vehicle section of the T2125 in April, or when a letter from the CRA asks for the log behind a claim already filed. Both lead to the same idea, which is to sit down and write the log now.
That idea has two versions, and the Tax Court treats them very differently. A log written from memory, with round distances and a business-use percentage picked first and trips filled in to match, is an estimate, and the Court has described it as one in almost those words. A log rebuilt trip by trip from records that were created at the time you drove, each trip pointing to the document that proves it, is evidence. It is weaker than a log kept as you went, but it can carry a claim, and it has.
This page is about building the second kind, and about being honest with yourself about where it runs out.
What the CRA asks for, and what it says when you do not have it
The CRA's motor vehicle records page calls an accurate logbook of business travel maintained for the entire year the best evidence of business use: for each business trip, the date, the destination, the purpose and the kilometres, plus the odometer at the start and end of the fiscal period. What a full log contains, field by field, is in its own guide.
"Best evidence" is a ranking, not a monopoly. The same page, describing what happens when a sample logbook fails, says business use for the rest of the year "would need to be determined based on an actual record of travel or alternative records". Alternative records are contemplated. They are also, by the CRA's own framing, second best, and the burden of making them persuasive is yours.
Two things follow before you start.
- A reconstruction is labelled as one. It is dated the day you prepared it, it says what it was built from, and it never pretends to be a log kept at the time. The difference between reconstruction and fabrication is not the format of the spreadsheet. It is whether every row is honestly described and traceable to a document.
- The result may be lower than what you claimed. A reconstruction that only counts trips you can evidence will usually find fewer business kilometres than a guess did. That is the point of it.
What each kind of record actually proves
Every business trip needs four facts: date, destination, purpose, distance. No single record supplies all four, so the method is to stack records until each trip has each fact from somewhere. This is the map.
| Record | Date | Destination | Purpose | Distance |
|---|---|---|---|---|
| Calendar entries created at the time | Yes | Often | Often | No |
| Invoices, quotes, work orders with a client address | Yes | Yes | Yes | No |
| Emails or texts arranging a visit or delivery | Yes | Sometimes | Yes | No |
| Phone location history (Google Maps Timeline export) | Yes | Yes | No | Yes, as driven |
| Fuel, parking, toll and supplier receipts | Yes | The place of purchase | Rarely | No |
| Card and bank statements | Yes | The merchant | Weak | No |
| Service, inspection and repair invoices | Yes | Not applicable | Not applicable | The odometer |
Read the table by column, because that is where the weak spots are.
Purpose comes almost entirely from business documents: the invoice for the job, the quote you sent, the email that says "see you at 10 on Tuesday". Location history can prove you were at an address. It cannot prove why, and purpose is the field that separates a business trip from a personal one.
Distance comes from location history where you have it, and from mapping where you do not. It never comes from memory.
Date and destination are usually the part you already have, which is why a reconstruction that stops there looks complete and is not.
A note on the records themselves. Export them as they are. Do not edit old calendar entries to add detail, even accurate detail, because an entry modified in March of the following year no longer proves what you knew at the time, and electronic records carry the date they were changed. If an entry needs context, the context goes in your reconstruction, in a notes column, in your words, dated today.
The method, step by step
1. Pin down the year's total kilometres first. Business use is business kilometres divided by total kilometres, and the total is the half people forget. It comes from two odometer readings, one at each end of the fiscal period. If you did not photograph them, the next section recovers them.
2. List every day you might have driven for business. Walk your calendar, your invoices and your sent email for the whole year and write down every date with a client visit, a site, a delivery, a supplier run or a meeting away from your regular place of work. One row per trip, with a source column naming the document.
3. Attach the destination. An address from the invoice or the client file, not "client site". Where the location history shows the visit, note that too; two independent records agreeing on the same date and place are stronger than either alone.
4. Attach the purpose from a business document. "Kitchen quote, invoice 2025-041" is a purpose with a source. If the only evidence of a trip's purpose is that you remember it, mark it as such or leave it out. Trips that rest on memory alone are where a reviewer starts pulling a reconstruction apart.
5. Compute the distance by mapping the route. See the section below. Chain the day's stops in the order you drove them.
6. Remove what is not business. Driving between home and a regular place of work is personal, even if you stopped to post a parcel on the way. If your home is the base of your business operations, trips from home to clients are business travel, which is the distinction that decided one of the cases below.
7. Reconcile against the total. Total minus business is your personal driving. Look at that number and ask whether it is believable for your household. If it says one car drove 900 personal kilometres in a year, the business figure above it will not hold.
8. Write the cover note. One paragraph at the top: when you prepared the reconstruction, why there was no contemporaneous log, which sources you used, how you computed distances and the odometer figures, and what you excluded. Keep every source document with it for the same retention period as the return.
Distance: mapped, not remembered
A distance you remember is a number you chose. A distance a mapping service computes for a stated route is a number anyone can check, which is the whole difference.
- Use the route you actually drove, starting from where you actually were. If Tuesday was home, client A, supplier, client B, home, map that loop once. It is shorter than three separate round trips from home, and a reviewer who remaps it will notice if you counted it as three.
- Where location history exists, prefer it for the day's distance, and use mapping to check it rather than replace it.
- Keep the route, not just the number: the addresses in order and the service you used. The point of mapped distance is that it can be reproduced.
A mapped route answers "how far is it from here to there". It does not answer "did you go". One of the cases below turned on exactly that gap.
Recovering the odometer you did not photograph
If you have no reading for January 1 and December 31 (or your fiscal year's first and last day), look for any document that recorded the odometer on a known date:
- Oil change, tire and service invoices, which usually record the odometer reading.
- Provincial inspection or emissions records.
- Dealer records for warranty and recall work, which you can request.
- Purchase, sale, trade-in or lease-return documents.
- Insurance or roadside-assistance claims.
Find the reading closest to each end of the year on both sides, and interpolate: assume the kilometres between two readings were driven evenly across the days between them. That assumption is imperfect, and your cover note should say you made it. Two readings a few weeks apart bracket the year end tightly; two readings six months apart do not, and the further apart they are, the more the result is an estimate.
Service records are also what the CRA itself reaches for. In one of the cases below, the auditor worked out the taxpayer's total kilometres from repair invoices that showed odometer readings, because no log and no year-end readings existed. In another, the Crown relied on maintenance invoices and the taxpayer showed at trial that the mileage printed on them was wrong. Check each reading against its neighbours before you rely on it.
A worked example
A self-employed contractor with a December year end, one vehicle, no log for 2025, and no odometer photos. Here is the total, rebuilt from four service records.
| Record | Date | Odometer |
|---|---|---|
| Provincial inspection | 2024-11-18 | 45,900 |
| Oil change | 2025-01-20 | 48,310 |
| Tire change | 2025-12-09 | 71,020 |
| Oil change | 2026-02-03 | 73,450 |
January 1, 2025. Between the inspection and the January oil change: 2,410 km over 63 days, about 38.25 km a day. January 1 is 44 days after November 18, so 45,900 + (38.25 × 44) ≈ 47,583.
December 31, 2025. Between the tire change and the February oil change: 2,430 km over 56 days, about 43.39 km a day. December 31 is 22 days after December 9, so 71,020 + (43.39 × 22) ≈ 71,975.
Total 2025 kilometres ≈ 71,975 − 47,583 = 24,392.
Then the trips. The contractor walks the year and keeps only trips with a document behind them.
| Source | Trips | Mapped km |
|---|---|---|
| Invoices and quotes with a job-site address, cross-checked against Timeline where it was on | 212 | 9,860 |
| Supplier receipts for materials picked up, chained into the day's route | 58 | 1,450 |
| Evidenced business kilometres | 270 | 11,310 |
| Trips remembered but with no document | Excluded | Not counted |
Business use = 11,310 ÷ 24,392 = 46.4%.
Two things to notice. First, the interpolation matters but does not dominate: if the total were 600 km higher or lower, business use would move to 45.3% or 47.5%. Second, the excluded row is doing real work. It is the reason the 46.4% can be defended line by line, and it is what separates this from a year-end guess. If the contractor originally filed 70%, the reconstruction does not support 70%, and the honest course is to say so rather than find more trips.
What the Tax Court has said
These are decisions of the Tax Court of Canada on vehicle claims without a contemporaneous log. Most were heard under the informal procedure, and section 18.28 of the Tax Court of Canada Act says an informal judgment is not a precedent for any other case. Treat them as illustrations of how evidence gets weighed, not as rules.
Year-end estimates are estimates. In Dore v. The Queen, 2004 TCC 638, the taxpayers stopped keeping mileage logs, kept partial logs for part of one year, and then estimated business use. Justice Bowie called it "simply a guesstimate based on partial information" and said estimates made at year end by subtracting a guessed personal share from the annual total "are only that, estimates. Generally, they tend to be generous to the estimator." The assessor had extrapolated from the partial logs that did exist, and the Court upheld her figure.
A log rebuilt from work records, but vague about where and why, did not help. In Glawdecki v. The Queen, 2010 TCC 650, an employee produced spreadsheets after the audit, built from his work time sheets and distances he worked out later with GPS software. He acknowledged he had not created them at the time. The Crown said they gave no specifics about where he travelled or why, and Justice Lamarre agreed that without a logbook created at the time, a taxpayer "has an evidentiary hurdle that is not easily overcome." Two details are worth knowing. The auditor had built her own estimate from repair invoices showing odometer readings, meal receipts marking office days, and a mapped commute. And the taxpayer's own reconstruction, once examined, showed most of his driving was commuting, which supported a lower employment share than the auditor had already allowed.
A reconstruction that contradicts itself is unreliable. In Platis v. The Queen, 2010 TCC 150, logbooks reconstructed at the CRA's request showed business mileage on weekends and statutory holidays, and on days the taxpayer was out of the country they showed the same 41 or 42 km as his drive from home to the office. He conceded in testimony that half his mileage was personal. Justice Angers found there was "no doubt" the logbooks were "reconstructed after the fact and are unreliable", and half the vehicle expenses stayed disallowed. The same judgment upheld gross negligence penalties on other expenses in that return.
Mapped distances for a documented routine were accepted, in part. In Morrissey v. The Queen, 2011 TCC 373, a newspaper distributor with no logbooks, whose vehicle claim the auditor had disallowed entirely, used Google Maps to show the kilometres she had to drive each day from her home, which the Court accepted as her place of business, to the depot and occasionally to checkpoints. The Crown pointed to maintenance invoices showing fewer kilometres, and at trial those invoices were shown to contain errors. Justice D'Auray accepted the calculations as reasonable given the evidence and the work, and allowed roughly three quarters of the amount claimed in each year. What carried it was a routine the Court believed, documented by the nature of the work, with distances anyone could check.
Mapped distances without proof of the trips were not. In Perron-Ali v. The Queen, 2021 TCC 6, an employee who had no log produced a Google Maps table of the distance from his home to every school in his board's territory. He did not go to all of them, and his own testimony suggested most of the claimed kilometres were driving to and from work. Deputy Judge Jorré found no basis for the claim: without a log or similar record, there was no way to separate deductible travel from commuting.
A log for a different year proves little about this one. In Martin v. The Queen, 2009 TCC 3, a standby-charge dispute, the appellants had no logs for the years in issue and offered logs prepared for later years as representative. Justice Margeson noted that while the Act does not require a log, when the use of a vehicle is in issue "such evidence is very important", and attached "very little weight" to books created after the fact for years that were not in issue.
And the door that stays open. In Malik v. The Queen, 2011 TCC 224, heard under the general procedure, Justice Hershfield observed that start-of-year and end-of-year mileage, set against the likely distances to known business destinations, "is not of the type generally preferred, compared to detailed mileage logs", but "it is, nonetheless, possible to reconstruct with some degree of accuracy some percentage of business use for an automobile", and that even an appointment book might help. Mr. Malik had none of it, so the Court could not go beyond the share the auditor had already allowed.
Read together, the pattern is consistent. What fails is a total guessed at year end, a reconstruction vague on destination and purpose, one that contradicts other evidence, and distance without proof the trip happened. What has succeeded is specific, checkable distance tied to business activity the Court found credible, and even then it recovered part of the claim rather than all of it.
What a reconstruction cannot fix
- It cannot be backdated. A reconstruction that presents itself as a log kept at the time is a false record, not a weak one. Date it the day you made it.
- It cannot invent a purpose. If the only reason a trip was business is that you say so, it is the weakest row in the file.
- It cannot rescue a percentage chosen first. Work from documents to a figure, never from a figure back to documents.
- It cannot survive contradiction. One trip on a day your card statement puts you in another city undermines every other row. Cross-check before anyone else does.
- It does not replace the base year. The simplified logbook needs a complete twelve-month log first, and a reconstruction is not one. More on that below.
Start the real log today
The cheapest fix is the one that begins now. Photograph the odometer today and write the date on it; that is the anchor for everything that follows. From today, log every business trip with date, destination, purpose and kilometres, as it happens. If you would rather not do it by hand, Claimable Drive, in beta for iPhone, records drives in the background and you swipe each one business or personal. Any method works if it captures the four fields, the purpose included.
There is a second reason to start now. The CRA's simplified logbook lets you keep a three-month sample in later years instead of a full log, but only once you have one complete twelve-month logbook to use as a base year. Every month you delay pushes that base year back. Kept properly, one full year now is what makes the following years lighter.
For the running habits around it (receipts, the per-vehicle split, and what goes on the T2125), see tracking vehicle expenses when you are self-employed. To see what your percentage is worth against actual costs, and why the per-kilometre rate is not a claim the self-employed can make, use the mileage calculator.
FAQ
Can I claim vehicle expenses if I did not keep a mileage log? You can claim, but you have to be able to support the business-use percentage. The CRA calls a full-year log the best evidence and contemplates alternative records. Without a log, expect the claim to be tested against whatever records you do have, and expect an auditor who finds too little to build their own estimate, which may be lower.
Is a reconstructed mileage log acceptable to the CRA? It is not automatically accepted or rejected. A reconstruction built trip by trip from records made at the time, with mapped distances and a clear note on how it was prepared, is evidence. One written from memory at year end is an estimate, and the Tax Court has said estimates of that kind tend to be generous to the person making them.
Can I use Google Maps Timeline as a mileage log? It supplies the date, the places you stopped, and the distance as driven, if Timeline was turned on. It does not record why you went, so it needs a business document beside each trip. Google now stores Timeline on the device, and on Android you can export it from Settings, Location, Location services, Timeline. Export it before you change phones.
How do I find my odometer reading for January 1 if I never wrote it down? Look for service, inspection, repair or dealer records dated on either side of January 1 and interpolate between them. The closer the two readings are to the date, the better the figure. Say in your notes that the reading is interpolated and from which documents.
I had a log and lost it. Is that different? It helps to be able to show the log existed, for example an old backup, a partial copy, or an app account. The CRA's record-keeping guidance asks that lost or damaged electronic records be reported and recreated within a reasonable time. The recreated version is still a reconstruction, and should be described as one.
What happens if the CRA does not accept my reconstruction? The vehicle claim is reduced to what the evidence supports, which can mean the auditor's own estimate. You can object and, if needed, appeal, but the burden of showing the reassessment is wrong stays with you, which is why the reconstruction should be built to survive a line-by-line check.
Sources
- CRA, Motor vehicle records, the logbook as "the best evidence", the four trip fields, the odometer requirement, and "alternative records".
- CRA Guide T4002, Self-employed Business, Professional, Commission, Farming, and Fishing Income, motor vehicle expenses on the T2125.
- Interpretation Bulletin IT-521R, Motor Vehicle Expenses Claimed by Self-Employed Individuals, archived, paragraph 8: expenses must be reasonable and supported, and a record of total and business distance kept with date, destination and distance for each trip.
- CRA, Your responsibilities and the requirements associated with records the law requires you to keep, records must be reliable, complete and supported by documents; lost electronic records are reported and recreated.
- Tax Court of Canada Act, section 18.28, informal procedure judgments are not precedents.
- Dore v. The Queen, 2004 TCC 638, paragraphs 20 and 21.
- Glawdecki v. The Queen, 2010 TCC 650, paragraphs 7 to 16.
- Platis v. The Queen, 2010 TCC 150, paragraph 17.
- Morrissey v. The Queen, 2011 TCC 373, paragraphs 8 to 13.
- Perron-Ali v. The Queen, 2021 TCC 6, paragraphs 36 to 45.
- Martin v. The Queen, 2009 TCC 3, paragraphs 165 and 166.
- Malik v. The Queen, 2011 TCC 224, paragraphs 24 to 26.
- Google Maps Help, Manage your Google Maps Timeline, on-device storage and exporting Timeline data.
Claimable Drive keeps this logbook for you.
An iPhone logbook, free for thirty drives a month, that records without being opened, suggests which drives are business, and shows the percent a self-employed Canadian actually deducts.
Get the free logbook