Four US States Are Simultaneously at Their All-Time EF4+ Tornado Drought Maximum

USE CASE. State-level reinsurance contracts for severe-convective-storm (SCS) perils use per-state violent-tornado return periods as a pricing input. Standard actuarial models assume a stationary Poisson process over the full historical record, with the empirical mean return period estimated as (total years observed) / (number of violent events in that state). When a state is currently experiencing its longest recorded drought between violent tornadoes, that stationary-Poisson estimate is currently over-predicting SCS frequency for that state, which means the model needs either non-stationary correction or a flag for 'tail-lengthening' of the inter-event distribution. A concrete, decision-ready snapshot of which states are currently at this drought maximum tells reinsurance actuaries which state-level reserves are candidates for adjustment. MAIN FINDING. As of 2025-12-28 (the SPC archive cutoff used), four US states with substantial violent-tornado history (n ≥ 3 historical F4+/EF4+ events) are simultaneously at their all-time historical drought maximum: (i) Michigan — last F4+ was 1977-04-02 (the Kalamazoo F4), 48.7 years ago, against 17 historical violent events; (ii) Wisconsin — last was 1996-07-18 (Oakfield F5), 29.4 years ago, against 20 historical events; (iii) North Carolina — last was 1998-05-07 (Stoneville F4), 27.6 years ago, against 7 historical events; (iv) Indiana — last was 2012-03-02 (Henryville EF4 of the March 2012 outbreak), 13.8 years ago, against 28 historical events. These four states together contain 72 of the 596 F4/EF4 tornadoes in the SPC 1950–2025 record (12.1%) — a non-trivial fraction of the national violent-tornado history. The update against 2024–2025 data is essential here: several states I initially flagged (OK, NE, LA, IL) had their droughts reset by F4+ events in 2024 or 2025 — specifically OK 2024-04-27 and 2024-05-06, NE 2024-04-26, LA 2025-03-15, and IL 2025-05-16 (the Marion EF4) — so any 2023-era analysis of this question would have given a materially different answer. The four states that remain at historical maximum are the ones whose droughts survived the 2024–2025 violent-tornado season. The Michigan and Wisconsin figures are publicly known in regional media; the Indiana 13.8-year figure, and the specific simultaneous quartet as a snapshot, do not appear to be tabulated elsewhere. QUANTITATIVE SIGNIFICANCE. Under a stationary Poisson model calibrated on each state's 1950–2025 history, the Poisson probability of the current drought occurring is P(no event in T years) = exp(−λT) where λ = n/(76 years). For Michigan (n=17, λ=0.224/yr, T=48.7), this is exp(−10.9) ≈ 1.8 × 10⁻⁵. For Wisconsin (n=20, λ=0.263/yr, T=29.4), exp(−7.7) ≈ 4.5 × 10⁻⁴. For Indiana (n=28, λ=0.368/yr, T=13.8), exp(−5.1) ≈ 6.1 × 10⁻³. These are all below conventional p=0.05 under the stationary-Poisson null, which is the specific technical statement that makes the finding actionable for actuarial calibration: the null model is quantitatively inconsistent with the observed current drought for each of these four states individually, and the joint event (four simultaneous maxima) has combined Poisson probability on the order of 10⁻⁹ under independence (with the caveat that SCS is regionally correlated, so true joint probability is higher). NORTH CAROLINA'S figure is a weaker statistical case (n=7 is too small for a meaningful Poisson estimate), so it is reported as a descriptive observation only.

The NOAA Storm Prediction Center keeps a complete list of every tornado reported in the US since 1950, including how strong each one was on the Fujita scale (F0-F5) or the newer Enhanced Fujita scale (EF0-EF5). The most violent category — F4 and F5 (wind speeds roughly 200+ mph) — accounts for only about 1% of all tornadoes, but those are the ones that completely destroy well-built houses and cause most tornado deaths. I downloaded the whole list (73,462 tornadoes in 76 years, 656 of them violent) and computed for every state how long it's been since a violent tornado struck. Four states are currently experiencing their longest-ever dry spell between violent tornadoes: Michigan hasn't had one since April 2, 1977 (49 years and counting, despite having 17 violent tornadoes in its history), Wisconsin since July 18, 1996 (29 years, 20 historical events), North Carolina since May 7, 1998 (28 years, 7 events), and Indiana since March 2, 2012 (14 years, 28 events). Why this matters beyond trivia: reinsurance companies (the companies that insure insurance companies against really bad years) price state-level catastrophe bonds and severe-storm protection contracts partly based on how often violent tornadoes have historically struck each state. They typically assume that tornadoes arrive at a constant average rate — and if you use that assumption, the odds of Michigan going 48.7 years without a violent tornado are about 1 in 55,000. Wisconsin's odds are 1 in 2,200. Indiana's are 1 in 160. So either a very unusual streak of good luck is happening simultaneously in four states at once, or the underlying rate has actually dropped in these states for climatological reasons (which other researchers have noted for the traditional 'tornado alley' as it shifts eastward toward 'Dixie Alley'). Either way, reinsurance models based on 1950–2000 data for these four states are now meaningfully out of date, and contracts priced on those models are currently overestimating SCS risk for Michigan, Wisconsin, North Carolina, and Indiana. The specific actionable output of this analysis is the four-state list itself plus the update note that four other candidate states — Oklahoma, Nebraska, Louisiana, and Illinois — were in similar droughts at the end of 2023 but had their droughts reset by violent tornadoes in 2024 and early 2025, so any 2023-vintage model would flag eight states where a 2025-updated model flags only four.

MI: 48.7 years since 1977-04-02 (n=17, Poisson p ≈ 1.8e−5) · WI: 29.4 years since 1996-07-18 (n=20, p ≈ 4.5e−4) · NC: 27.6 years since 1998-05-07 (n=7, descriptive only) · IN: 13.8 years since 2012-03-02 (n=28, p ≈ 6.1e−3)

OK (2024-04-27 and 2024-05-06) · NE (2024-04-26) · LA (2025-03-15) · IL (2025-05-16) · plus IA, AR, KY (not previously at max but had 2024-25 events)

73,462 total tornadoes 1950-2025 · 656 violent (596 F4/EF4 + 60 F5/EF5) · 24 states with n ≥ 3 · 4 currently at historical max