Before You Amend a Cannabis Return, Run This 10-Minute Feasibility Check

This one is for the boutique cultivator who has been told — by a refund-claim marketer, a podcast, a peer at a conference — that you should amend your prior-year returns to claw back the §280E tax you paid when everything was still Schedule I. Maybe you should. Maybe you absolutely should not. The pitch never includes the part where amending can cost more than it returns, so before you sign anything, run this check. Five questions decide it, and you can take the first pass yourself right now — or have me run them against your actual numbers, which is exactly what the $375 audit does.

The five questions

1. What is the basis of the refund? This is the one that matters most. A claim resting on the theory that "medical cannabis was always really Schedule III, so my prior §280E disallowance was wrong" is the contested position — it depends on giving the 2026 rescheduling retroactive effect, and the IRS is actively pushing back on cannabis refund claims built on it. Retroactive §280E relief has not been granted; at most it has been encouraged, and the Service's posture is adverse. A claim resting instead on a §471-11 or §471(c) correction — indirect production costs you were entitled to capitalize into COGS and did not, a full-absorption computation you never ran — is a different animal. That is ordinary tax law you underapplied, not a bet on how a scheduling order gets interpreted, and it is far more defensible.

2. What is the dollar amount? Run the real math: the refund, minus the cost of the engagement, minus an audit-exposure premium for opening the year. If the number that survives is small, the calculus fails even when the position is sound. A defensible claim for trivial dollars is still not worth filing.

3. How old is the year? Refund claims are time-limited. Under §6511, a claim is generally due within three years of filing the return or two years of paying the tax, whichever is later. Years outside that window are out of reach no matter how strong the merits — and no marketer's enthusiasm changes the statute.

4. Are the records there? An amended return has to be supported by workpapers — the schedules that tie a corrected COGS to your books, your Metrc, and your POS. If the year's records would have to be reconstructed from scratch, that reconstruction is real work and real cost, and it belongs in the feasibility math before you commit.

5. What is the audit exposure? Amending a year opens that year to examination. If the year carries unrelated exposure — cash-handling and Form 8300 gaps, payroll-tax trouble, a shaky allocation method — a refund claim can invite a look that costs you more than the refund returns. You do not want to open a door you are not ready to have the IRS walk through.

The decision, in four lines

Put the answers together:

• Defensible basis, meaningful dollars, inside the §6511 window, records that exist, low unrelated exposure → amend.
• Contested retroactive-§280E basis → do not amend on the theory. Wait for settled guidance, or pivot to the going-forward method, where the money is sure.
• Small dollars or heavy reconstruction → usually not worth it; put the energy into the 2026 going-forward win instead.
• High unrelated exposure → fix the exposure before you amend anything.

The going-forward alternative is usually the better play

Here is what the refund-claim pitch leaves out: the 2026 return is where the sure money lives, with none of the audit risk of reopening old years. A corrected medical-vs-adult-use allocation, full-absorption COGS, and a properly chosen §471(c) method reduce your ongoing liability going forward, and they do it without betting on a contested theory. So the amended-return question is one you should answer after the going-forward method is set — because doing that method work is what surfaces the legitimate prior-year corrections actually worth amending for. It is also worth remembering that even the going-forward medical position rests on the 2026 rescheduling order, and that order is under active challenge — a stay was requested in the D.C. Circuit in July 2026 and, as of now, has not been granted, so the order stands, but you want positions built to stay defensible if the ground shifts. And treat the precise 2026 transition-year timing as something to confirm against current IRS guidance before filing, rather than assume.

If you do amend, do it supported

When a claim clears the check, file it supported rather than exposed. A §471 method memo, plus the right disclosure — Form 8275, or Form 8275-R if the position is contrary to a Treasury regulation — documents the position (I cover 8275-R as disclosure, not a shield, in a companion guide). If a year is near the edge of its §6511 window while the law is still unsettled, a protective refund claim can preserve the window without betting the outcome — but understand it for what it is: a window-preserver, not a guarantee that the refund is coming. And the amended return is signed under my Circular 230 obligations as your Enrolled Agent — which is another way of saying I do not put my signature on a claim that has not cleared this check.

The quick checklist

1. Identify the refund basis — a §471 correction, or a retroactive-§280E theory?
2. Estimate the dollars net of engagement cost and audit exposure.
3. Confirm the year is inside the §6511 window (three years from filing / two years from payment).
4. Confirm the records exist or can be reconstructed — and price the reconstruction.
5. Assess the year's unrelated audit exposure.
6. Decide: amend now, set the going-forward method first, or pass.

What a refund claim actually costs

A refund claim is not free even when it works — filing it opens the year to examination, and a claim that fails can draw accuracy-related penalties under §6662 on top of the disallowance. That is the whole reason the feasibility check exists: to confirm the expected value is positive before you file, not to talk yourself into a refund that was never coming. I will never tell you a refund is guaranteed — no honest EA can — but I can tell you, against your real numbers, whether the claim is worth making.