Mirch v. Commissioner: the on-call hours that sank an STR loss

By Max Medvedev · 6 Aug 2026

The loss that left the rule standing

The owners of a short-term rental claimed roughly 920 participation hours and deducted the loss against their W-2 income. The Tax Court agreed the property met the seven-day test, then disallowed the entire loss. 744.5 of those hours were on-call time, and credible participation came in under 100 hours.

That is the shape of nearly every recent short-term rental loss in the Tax Court. The seven-day route works. What fails is the record. Mirch v. Commissioner, T.C. Memo 2025-128, is the clearest current example of the difference.

What happened

The property cleared the front gate. Its average stay was seven days or less, which takes the activity out of the "rental activity" definition in Reg. §1.469-1T(e)(3)(ii)(A). The owners then had to clear the second gate — material participation under Reg. §1.469-5T(a) — and offered an activity log as proof.

The log claimed about 920 hours for the year. Its blocks each carried a distinct problem:

• 744.5 hours of site management and on-call time — the largest block by far, and time spent available rather than time spent working.
• 168 hours of cleaning — claimed in a year whose return also deducted payments to professional cleaners.
• A flat seven hours per turnover, applied identically whether the stay ran one night or fourteen.
• Undated entries, assembled after the year closed rather than as the work happened.

Behind all of it sat a W-2 job — the first thing an examiner establishes, because it sets the ceiling on what the rest of the claim can plausibly fit around.

What the court held

Two holdings, and the order is the point.

First, the classification survived. The court accepted that the property met the short-term rental definition. The deduction did not fail because the rule is narrow, disputed, or newly under attack.

Second, that classification only removes the automatic passive label. Material participation still had to be proven, and on this record it was not. Once the unsupportable categories came out, credible participation stood under 100 hours — beneath the lowest hour test in the regulation — and the entire loss was disallowed.

The standard the log had to meet is not harsh on its face. Reg. §1.469-5T(f)(4) allows participation to be established "by any reasonable means" and does not require daily time reports. The gap between that text and these outcomes is the whole story of this case library: the regulation would let you reconstruct, and the courts punish reconstruction. Moss v. Commissioner, 135 T.C. 365 (2010), said it first — the regulations "do not allow a post-event ballpark guesstimate."

The tell

The deciding behaviour was not the size of the total. It was that the log described someone who was available rather than someone who worked — and that it could be checked against the rest of the return.

Roughly four-fifths of the claimed year was on-call time. Being reachable while a guest is in the house is real life, and it is not participation; Moss had settled that fifteen years earlier. Once that block came out, there was very little left to argue about.

The 168 cleaning hours are the second half of the tell, and the more instructive half. They did not fail because cleaning does not count — cleaning you do yourself counts fully. They failed because the same return deducted professional cleaners for that work. A log that contradicts a line on the return it supports discredits itself, and it invites a court to discard the entries nobody has checked yet. That is the real cost of over-claiming: 800 defensible hours are worth more than 1,200 inflated ones, because the inflated ones take the good hours down with them.

The flat seven hours per turnover is the same failure in miniature. Identical durations for tasks of obviously different size are the signature of a number that was chosen rather than recorded.

What this means for your records

Five checks fall out of this case, and they are what a year's log can be tested against before anything is filed:

Every one of those is a property of when and how the entry was made, not of how the year was summarised afterwards. That is why the fix is structural rather than editorial: an entry written the day the work happened, sized to the task, with the guest message thread, invoice, receipt, or photo attached to it, answers all five before anyone asks. The same entries assembled in April cannot, no matter how honest the underlying year was.

The four-gate analysis and the hour race behind it are covered in the hub for this cluster, how to prove your short-term rental hours, and the average-stay math that Mirch passed is in the seven-day test.

See where your year lands

Run your reservation mix and hours through the estimator — it applies the seven participation tests in Reg. §1.469-5T(a), shows which one you pass or miss, and cites the rule behind every number.