Will there be between 14 and 16 earthquakes of magnitude 7.0 or higher worldwide in 2026?

TL;DR

The forecast is 69.5% Yes and 30.5% No for a worldwide 2026 total of 14–16 earthquakes of magnitude 7.0 or higher. The USGS catalog records 13 qualifying events at the November 4 cutoff, leaving a target of one to three additional events with classifications unchanged. The historical rate supports that target; the main threats are a cluster that takes the total above 16, no further qualifying events, and revisions across the magnitude threshold.

Context

This market counts distinct earthquakes worldwide with a preferred USGS magnitude of at least 7.0. Qualifying aftershocks and earthquakes of magnitude 8 or higher count too. The designated USGS significant-earthquake page also contains smaller earthquakes, so its total number of entries is not the answer to the market.

The information cutoff is November 4, 2026, at 00:00 UTC. Approximately 58.2 occurrence-days remain until December 31 at 11:59 PM ET. The reporting allowance through January 7, 2027, lets an earthquake that occurred within the window appear later in the resolution record. It does not extend the occurrence window.

Evidence

The historical backbone is a global rate close to 15 qualifying earthquakes per year. The current USGS catalog contains 541 events over the 36 complete years from 1990 through 2025, giving an annual mean of 15.0278. The annual sample variance is 13.8563492, which does not show unusually wide year-to-year variation relative to that mean. These figures use one consistent, current catalog vintage and the preferred magnitude for each event (USGS event-service documentation).

The full annual history is below. All values are worldwide counts of earthquakes with preferred magnitude at least 7.0 in the USGS catalog reference class.

Year	Count	Year	Count	Year	Count
1990	18	2002	13	2014	12
1991	17	2003	15	2015	19
1992	13	2004	16	2016	16
1993	12	2005	11	2017	7
1994	13	2006	11	2018	17
1995	20	2007	18	2019	10
1996	15	2008	12	2020	9
1997	16	2009	17	2021	19
1998	12	2010	24	2022	11
1999	18	2011	20	2023	19
2000	15	2012	16	2024	10
2001	15	2013	19	2025	16

The more recent 2000–2025 period contains 387 events and averages 14.884615 per year. That is close to the longer reference period. The evidence does not establish a persistent trend that calls for a different global rate. Nor does it show that a year with 13 events by early November must accelerate or slow down to reach an average. Historical earlier-year activity had little demonstrated relationship with the remaining count.

Applied to the remaining occurrence window, the longer historical rate implies approximately 2.3947 additional earthquakes. That is an expected count, not a prediction of a particular integer. Its relevance is simple: the target requires one, two, or three more events, and the historical pace lies inside that interval.

Calendar-matched history gives a useful check. Across 1990–2025, the comparable late-year windows averaged 2.611 events, with variance 2.644. Those windows also show why the tails matter: the 2004 window contained eight qualifying earthquakes, while 2020 contained none. The sample does not establish a dependable seasonal uplift. Its role is to test whether the broad annual rate gives a reasonable picture of the remaining period, not to imply that November and December follow a special rule (USGS catalog data).

The current starting point is equally important. The 2026 USGS archive records the following 13 reviewed qualifying events. Dates are UTC and magnitudes are the current preferred values.

Date, 2026	Location	Magnitude
February 22	Malaysia	7.1
March 24	Tonga	7.5
March 30	Vanuatu	7.3
April 1	Indonesia	7.4
April 20	Japan	7.4
June 7	Philippines	7.8
June 24	Venezuela, first event	7.2
June 24	Venezuela, second event	7.5
July 17	Mexico	7.3
August 10	Colombia	7.4
August 14	Indonesia	7.8
October 9	Panama	7.7
October 22	China/Tibet, near Nagqu	7.0

The two Venezuelan earthquakes have separate catalog records. Their close timing is not a reason to merge them. The market counts qualifying earthquakes, not only independent mainshocks.

Short-lived clustering is real. Historical gaps of a day or less occur more often than a simple constant-rate process implies, and roughly two-month counts show modest extra variability. But the effect is not evidence of a lasting global surge. A USGS-hosted study of global clustering supports a stable global baseline with localized aftershock sequences. I read this as a reason to respect the tails, not to assume that recent large earthquakes predict worldwide acceleration.

The local signals are limited. The Panama aftershock forecast, issued November 2 at 16:54 UTC, gives a 2% chance of at least one M7-or-higher aftershock within a month and 5% within a year. Those are local probabilities. Historical worldwide counts already include aftershocks, so simply adding them to the global baseline would double-count part of the risk.

Magnitude classification is the other meaningful edge. The Nagqu event remains reviewed at M7.0. Its finite-fault product gives a derived magnitude of 6.94834, displayed as 6.9. That discrepancy matters because the event sits on the market threshold. It does not mean the preferred catalog magnitude has already changed.

A separate apparent extra event does not qualify. The October 25 Bonin Islands earthquake had a Japanese headline magnitude of 7.0, but JMA's centroid-moment-tensor page reports Mw6.5, agreeing with the USGS preferred M6.5. This is a magnitude-scale difference, not established evidence of a missing qualifying earthquake.

What's non-obvious

This is a range forecast, not an “at least 14” forecast. More earthquakes do not always help Yes. Starting from 13, the first additional event enters the target range, but the fourth leaves it. Another quiet week can therefore strengthen Yes by reducing the risk of overshooting 16, even though it also leaves less time to reach 14. That logic does not require earthquakes to be overdue or the annual total to converge on its historical average.

The other easy mistake is to confuse a physical estimate with the catalog value that controls the count. Nagqu's finite-fault estimate below 7.0 is relevant uncertainty, but the reviewed preferred magnitude remains 7.0. Conversely, a headline magnitude of 7.0 on another scale does not automatically add an event. The forecast follows the qualifying catalog count while recognizing that threshold classifications can change.

Uncertainties

The largest uncertainty is ordinary future occurrence. If the current 13-event count holds, no additional earthquake leaves the final total below the range, while four or more take it above. A concentrated sequence is the clearest route to overshooting. The global historical rate is useful, but it does not identify when or where the remaining events will happen.

Revision risk is less well measured. Five excluded 2026 earthquakes currently have preferred magnitude 6.9: May 25, June 24, August 15, October 16, and October 24. They create possible upward-revision exposure, just as Nagqu creates downward-revision exposure. Their proximity to the threshold is not evidence that revisions are imminent. A historical study of changes between first publication and final preferred magnitude would help quantify this risk; that evidence is not established here.

Historical comparability also has limits. The static USGS statistics table differs from the current event catalog in some years. For example, it lists 16 for 2001 versus 15 in the current catalog, and 14 for 2012 versus 16. The event-level reasons remain unresolved. Using a consistent current catalog avoids mixing incompatible histories, but it cannot reconstruct what each past year-end count looked like at the time.

These limits do not change the forecast. They explain why the historical fit and the present count support Yes without making it certain. The most useful new information will be the next qualifying event, any revision to a threshold event, and the amount of occurrence time remaining.