A Roth conversion counts for the tax year in which the money leaves the traditional IRA (IRS Pub. 590-A). There is no April catch-up and, since the TCJA repealed recharacterization in 2018, no undo. So the week between December 31 and January 1 is a real boundary: the same dollars, converted seven days apart, land in different tax years, different Medicare lookback years, different five-year-clock vintages, and different estimated-tax quarters. Each system favors one side of the boundary. This guide prices all five.
1. Bracket headroom: December is measured, January is forecast
By mid-December you know the year’s income within a few dollars: the last dividend has paid, the bonus has cleared or hasn’t, mutual fund capital gain distributions have posted. A December conversion can fill to the top of the 22% bracket ($105,700 taxable, single; $211,400 MFJ in 2026) with precision.
A January conversion is sized against a forecast of the next twelve months. A $10,000 income surprise in either direction is a $2,200 error at 22%, and worse if it pushes the top of the conversion across $105,700 into the 24% bracket. You can correct course with a second conversion in December, but you cannot pull back an overshoot. Forecast risk is asymmetric, which is why January conversions should be sized conservatively and topped up in December, not the reverse.
2. The five-year clock: December is 11 months shorter
Each conversion’s five-taxable-year recapture clock runs from January 1 of the conversion year, not from the transaction date. Convert December 15, 2026 and the clock expires January 1, 2031: an effective lock of 4 years and 17 days. Convert January 15, 2027 and it expires January 1, 2032. One month of delay adds eleven months of lock. If you are under 54.5 (where the full clock binds; the five-year rule guide derives that crossover), December wins this system outright.
3. The IRMAA lookback: January delays the premium hit one year
Medicare premiums for a year are set by MAGI from two years earlier (42 CFR 418.2135). A December 2026 conversion raises 2026 MAGI and shows up in 2028 premiums; a January 2027 conversion shows up in 2029. Crossing the first IRMAA tier costs $1,148.40 per person per year in 2026 ($2,296.80 for a couple both on Medicare), so for anyone who will be on Medicare in the earlier landing year, one week of calendar delay moves a four-figure premium charge one year further out.
The boundary case is sharper. Premiums start at 65, so the lookback first reaches the year you turn 63. Convert in December of the year you turn 62 and the conversion never enters the lookback window at all; wait until January and it lands squarely in your first premium year. For that one cohort, December is not an optimization, it is an exemption. The mechanics are in how to avoid IRMAA.
4. Estimated-tax float: January defers the cash roughly a year
Federal tax on a December 2026 conversion is due with the 2026 return, April 15, 2027, about 3.5 months of float (assuming your withholding or equal estimates already meet the §6654 safe harbor: 100% of prior-year tax, 110% if prior-year AGI exceeded $150,000). The same conversion executed January 2027 owes its tax April 15, 2028: roughly 15 months. A $50,000 conversion sitting inside the 22% bracket owes $11,000; twelve extra months of float on $11,000 at a 4% money-market yield is $440. Small next to the IRMAA and bracket numbers, but it is the one system that always favors January.
5. The pro rata denominator: December is nearly known
If your IRAs carry after-tax basis, the taxable fraction of a conversion depends on the December 31 value of all your traditional IRAs. A December conversion’s denominator is nearly final; a January conversion’s fraction floats on eleven more months of market drift before it is knowable (the pro rata guide computes a $444 retroactive shift from a single rally). No basis, no effect.
The scorecard
| System | December (year N) | January (year N+1) |
|---|---|---|
| Bracket headroom | measured; fill within dollars | forecast; size conservatively |
| Five-year clock | ~11 months shorter lock | full five calendar years |
| IRMAA lookback | premiums hit in year N+2 | pushed to year N+3 |
| Estimated-tax float | ~3.5 months | ~15 months with safe harbor |
| Pro rata denominator | nearly known | 11 months of drift |
Three of five favor December, and the two that favor January are worth about $440 of float plus, for Medicare-age converters, a one-year premium deferral that is a timing shift rather than a saving (the surcharge lasts one year either way). The exceptions where January genuinely wins: your current year’s bracket is already full (January is simply the next headroom available), or you are 63 or 64 and the deferral pushes a tier crossing past a year you particularly need it out of, such as the year a spouse’s filing status changes.
One operational note for December: the conversion must leave the traditional IRA by December 31, and custodians publish year-end processing cutoffs. Same-custodian electronic conversions complete in a day; transfers that cross custodians or move securities in kind can take weeks. A conversion planned for December 28 and settled January 3 is a January conversion, with every consequence in the table.
Open the calculator
to see both calendars: each projection year is a calendar-year conversion, so modeling a conversion this year versus next shows the bracket, IRMAA, and ACA consequences of the week between them.