How to Prove Short-Term Rental Hours: 651 Claimed, 84 That Count.

A surgeon and her husband bought a cabin in Heber in March, listed it on Airbnb, and paid $9,000 for a cost segregation study that turned the first year into a $118,000 loss. She called in October with a spreadsheet and one question: she has 651 hours on the property, so she is fine, right? She has 84. How to prove short-term rental hours is where the short-term rental loophole usually falls apart, and the gap is rarely about effort.

Every article on this strategy ends with the same sentence: keep a contemporaneous log. That is good advice and it is not the law. Here is the standard, from Temp. Reg. §1.469-5T(f)(4):

The extent of an individual's participation in an activity may be established by any reasonable means. Contemporaneous daily time reports, logs, or similar documents are not required if the extent of such participation may be established by other reasonable means.

The regulation then describes those other means: the services performed over a period and the approximate hours spent, drawn from appointment books, calendars, or narrative summaries. The format is open. A message thread, a calendar, and a stack of hardware store receipts can carry the day without a spreadsheet.

What the regulation gives you in format it takes back in substance, because the burden is yours. In Goshorn v. Commissioner, T.C. Memo. 1993-578, the Tax Court said it was not bound to accept a taxpayer's unverified, undocumented testimony, including a post-event ballpark guesstimate. I still tell clients to log as they go, not because §1.469-5T demands it, but because a note made the day of is the cheapest corroboration there is.

Mirch is the case to read, because the owners kept a log and still lost. The court accepted that the Reno property was a short-term rental, the easy half of the test, then looked at the hours. The log claimed eight hours of site management for each of the 93 nights the place was rented, 744 hours, roughly four fifths of everything claimed, and defined site management as being on call for guests, repairs, supplies, Wi-Fi, cable, and snow removal. The court allowed none of it.

That was not new law. In Moss v. Commissioner, 135 T.C. 365 (2010), on-call hours did not count toward the 750-hour real estate professional test under §469(c)(7)(B)(ii), because the landlord performed no actual work during them. The cleaning hours failed for a related reason: one flat estimate covered every stay, whether the guest was there one night or fourteen, which reads as an allocation rather than a record.

If your hours come out of a per-night or per-stay multiplier, you do not have 744 hours. You have one assumption repeated 93 times, and an examiner will say so.

Two more categories go to zero before you reach the 100-hour line. The first is investor work. Treas. Reg. §1.469-5T(f)(2)(ii) excludes work done in your capacity as an investor unless you are directly involved in day-to-day operations, and its examples include studying financial statements or reports on operations and monitoring in a non-managerial capacity. Reading your own profit and loss lands there.

The second is drive time. Nothing in §1.469-5T addresses it and I have not seen a court credit it on a short-term rental, so I leave it out and treat it as upside. A position that needs travel hours to clear 100 is one you cannot defend.

The 100-hour test in §1.469-5T(a)(3) has two halves and owners forget the second. Your participation also has to be at least as much as any other individual's, including people you pay, and the comparison runs person by person rather than in total. A cleaner at 96 hours and a handyman at 40 do not combine into 136. You have to match the cleaner, not the crew, and the test reads not less than, so a tie is enough. Proving what the cleaner did is your job too. Pohoski v. Commissioner, T.C. Memo. 1998-17, which the IRS passive activity loss audit guide lists as a no-material-participation case, involved two Hawaii condominiums run through a rental agent whose hours never made it into the record.

Two things make that easier than it sounds. Under §469(h)(5) your spouse's participation counts as yours, with no ownership interest and no joint return required, so a spouse who handles the messaging adds to your side of the comparison instead of the other. And the 500-hour test in §1.469-5T(a)(1) has no comparison in it at all, which is why an owner who self-manages a busy property is often better off aiming there. The facts-and-circumstances test people reach for as a backstop is not available: §1.469-5T(b)(2)(iii) shuts it off at 100 hours or less.

Hours are counted by calendar year, so December 31 closes the 2026 count and there are about ten weeks left. Four things are worth doing with them:

• –Export the platform message threads and your calendar now, while the timestamps are retrievable, and build the hours from them, not from memory.
• –Ask your cleaner or co-host for an hours total in writing. If theirs beats yours, you know in October instead of April, with time to take over the turnovers.
• –Do the next repair, the restock, and the listing photos yourself, time them, and keep the receipt. Real work in November counts; a reconstructed August does not.
• –If you cannot close the gap honestly, plan around it. A suspended loss is not a lost loss: it sits on Form 8582 until the activity produces passive income or you sell.

That last point is the honest verdict. Missing the test costs you the year, not the deduction, and on a $118,000 loss the year is worth about $34,887 at a 2026 rate you can see against a future rate you cannot. The year you buy and run a cost segregation study is the year the deduction is biggest, which is why the hours deserve attention in October and not at filing.