Every FOMC rate decision since 1990

Fourteen cycles

A cycle starts at the first move that reverses the direction of the previous one and ends at the last move in that direction; the holds inside it are counted. Fourteen cycles since 1990, seven hiking and seven cutting. The current one, opened on 16 September 2026, is the fourteenth: one move so far.

Two cycles stand apart. The 1990–92 cutting cycle, 18 moves for -525 basis points, was carried out with 14 of its moves outside scheduled meetings and none of them announced. The 2022–23 hiking cycle, 11 moves for +525 points in 497 days with a single pause, is the largest in the register; the 2004–06 cycle took 17 moves and 729 days for +425 points, never more than 25 at a time.

Who dissented, and which way

The FOMC publishes a roll call. Of the 292 decisions with a recorded vote, 181 were unanimous and 111 had at least one dissent: 81 with one dissenter, 22 with two, 6 with three, 2 with four. The two four-dissent votes are 2 October 1990, 7–4, and 29 April 2026, 8–4. The register reads the side of each dissenter from the statement: hawkish when they wanted a higher rate than the decision, dovish when they wanted a lower one, mixed when both sides dissented at the same meeting, other when the objection was about guidance or asset purchases rather than the rate.

The chart below plots every split vote on the path of the target rate. Pick a chair or a side; hover a marker for the date, the vote and the names.

The pattern by chair is the one the chart shows. Ben Bernanke faced a dissent at 41 of his 66 meetings, 62%, most of them in 2011–13 over forward guidance and asset purchases, when the target itself did not move for seven years. Alan Greenspan, 148 meetings, 24%. Janet Yellen, 32 meetings, 47%. Jerome Powell, 67 meetings, 27%, with a run of nine dissents in the last fourteen meetings of his chairmanship and the 8–4 vote of April 2026. Kevin Warsh has chaired 3 meetings, one of them with three dissenters, the hawks of July 2026, before the unanimous 12–0 hike of September.

What this does not prove

A dissent is a vote, not a forecast: the register does not say whether the dissenters were right, and a hawkish dissent in 2011 concerned the phrase “exceptionally low levels for an extended period”, not a rate. The count of dissents is not comparable across eras without care: the Committee had between eight and twelve voting members depending on Board vacancies, so 7–4 in 1990 and 8–4 in 2026 are the same number of dissenters against a different total. The side of a dissent is read from the statement wording by rule; forty-three dissents are classed as other because the wording does not name a rate, and the CSV keeps the sentence for each one. Before February 1994 the register dates changes by the Desk’s implementation, which is not the day the Committee decided. The regime column is the eco3min classifier’s reading of the month, with degraded inputs before 2003, and the 10-year column measures two closes, not the effect of the decision.

Method, and the register itself

Primary sources. The Federal Reserve’s “Open market operations” tables for every change of the target since 1990, 110 rows; the FOMC historical materials by year (1990–2020) and the meeting calendars (2021–2026) for the list of meetings and conference calls; the FOMC statements for the vote from March 2002, the minutes from 1993 to January 2002, the Records of Policy Actions for 1990–1992; the June 2026 minutes for the names behind the new statement format. FRED DGS10 via eco3min for the 10-year yield, the eco3min regime classifier for the monthly regime. Retrieved 19 September 2026.

Rules. One row per scheduled meeting, plus each unscheduled meeting or conference call that changed the target, plus each change implemented between meetings. The 15 September 2003 special meeting and the cancelled meeting of March 2020 are not rows. A change dated by the Desk on the day after a meeting is attached to that meeting. The target before 13 July 1990 is 8.25%%, read from the first entry of the table. Direction and size are those of the target, then of the upper bound of the range. A cycle is a run of moves in the same direction; a hold is measured from the last move of a cycle to the first move of the next. The vote is the count of names for and against on the policy action; the side of a dissent is read from the sentence that carries the preference. Regime: month of the decision in the classifier’s history file. Twenty-four rows have no vote: the changes implemented between meetings in 1990–1992, the conference calls of 1990–91, 1994, 1998 and 2001, and five meetings whose minutes are incomplete online (1992–93 and November 1999).

Questions about the register

How many times has the Fed changed rates since 1990?

110 times between July 1990 and September 2026: 52 hikes and 58 cuts, over 316 decisions. The other 206 decisions left the target unchanged.

What was the most divided FOMC vote?

Two votes with four dissenters: 2 October 1990, 7–4, and 29 April 2026, 8–4. In April 2026 one dissenter wanted a cut and three objected to the easing bias in the statement, so the register classes it as mixed.

How often does the FOMC vote unanimously?

181 of the 292 recorded votes, 62%. The share varies by chair: 62% of Ben Bernanke’s meetings had a dissent, 24% of Alan Greenspan’s, 47% of Janet Yellen’s, 27% of Jerome Powell’s.

What is the Fed’s longest hold?

2,556 days at 0–0.25%, from the cut of 16 December 2008 to the hike of 16 December 2015, 55 consecutive meetings without a change. The second longest is 731 days at 0–0.25% again, from March 2020 to March 2022.

What is the Fed’s latest rate decision?

16 September 2026: a 25 basis point hike to 3.75–4.00%, by a 12–0 vote, the first hike since July 2023 and the first move up 280 days after the last cut of 10 December 2025.

Read next on the site: the next FOMC meeting and the time of the decision, every ECB rate decision since 1999, the fed funds rate as the central signal of monetary policy, a typology of Fed hiking and cutting cycles since 1954, the Fed’s track record against the Taylor rule, every Fed cut into a rising market since 1984, Fed vs ECB, two mandates compared and the effective fed funds rate as a monthly dataset. The wider frame is the central banks and monetary policy sub-pillar of the monetary regimes and rate cycles pillar.

Last updated — 19 September 2026