Will 2026 be the hottest year on record?

TL;DR

The forecast is 73.1% Yes and 26.9% No for 2026 setting or tying NASA’s annual temperature record at its first qualifying publication. Strong summer Pacific warming supports the late-year catch-up needed to overcome the cooler start to the year. The decisive issue is whether that global warming response arrives within 2026: repeating August’s temperature through December leaves the annual average narrowly short.

Context

This question is about one numerical ranking, not whether every climate agency declares a record. The resolution target is NASA’s annual Global Land-Ocean Temperature Index, specifically the displayed No_Smoothing value. Its current history spans 1880–2025, and the highest annual value is 2024’s 1.29°C above the 1951–1980 baseline. The recent values for 2023 and 2025 are 1.17°C and 1.19°C, respectively (NASA annual index). A numerical tie for first counts as Yes. Later revisions do not reopen the result.

As of October 6, 2026, NASA’s monthly observations extend through August. September has ended, so other observations already provide evidence about it, but NASA’s September release is scheduled for October 9 (NASA release schedule). The forecast therefore rests on a known early-year deficit, observed late-summer warmth, a strong developing El Niño, and September evidence that tempers the simplest extrapolation.

Evidence

The historical starting point is that records are common in the recent warming climate, but they are not the default outcome every year. In the current NASA annual series, 17 of the 46 years from 1980–2025 set or tied the then-existing record, about 37%. Over 1997–2025, 11 of 29 years did so, about 38%. Those years were 1997, 1998, 2002, 2005, 2010, 2014, 2015, 2016, 2020, 2023 and 2024. These are retrospective counts using today’s historical values, not an audit of first-publication rankings (NASA annual index).

The current conditions are more informative than that unconditional frequency. Strong developing-El-Niño years have shown substantial late-year warming. September–December temperatures exceeded the June–August average by approximately 0.159°C in 1997, 0.254°C in 2015 and 0.241°C in 2023. Those analogues establish that the required catch-up is physically and historically credible. Three selected events do not establish certainty (NASA monthly history).

The full January–August comparison shows where the deficit came from. All figures below are anomalies in °C above 1951–1980, from the same current NASA table (NASA monthly observations).

Month	2026	2024
January	1.09	1.25
February	1.25	1.44
March	1.32	1.40
April	1.17	1.32
May	1.13	1.16
June	1.18	1.23
July	1.25	1.20
August	1.40	1.30

January–August 2026 averages 1.22375°C, compared with 1.28750°C for the same months of 2024. The deficit is 0.06375°C. July and August have moved ahead of the record year, but that does not erase the cooler preceding months (NASA monthly table).

The threshold arithmetic explains why the forecast favors Yes without treating it as settled. The eight observed months total 9.79°C-months. An underlying annual average of approximately 1.285°C would ordinarily display as 1.29°C and tie the current record. On that basis, September–December needs to average 1.4075°C. Reaching an unrounded annual average of 1.29°C would instead require 1.4225°C over the remaining months. These are approximate boundaries because the published monthly inputs are themselves rounded (NASA temperature tables).

August persistence alone is not enough. Four more months at August’s 1.40°C would produce a 1.2825°C annual average, narrowly short of the approximate tie boundary. Repeating September–December 2023’s sequence of 1.48, 1.35, 1.40 and 1.36°C would likewise yield 1.2816667°C for 2026. A favorable-looking late-year comparison can still miss because the starting deficit matters (NASA monthly history).

The strongest positive evidence is the Pacific warming already in place. Conventional weekly Niño-3.4 anomalies rose through the reported readings of 1.3°C on June 3, 2.4°C on July 29, 2.5°C on August 5, 2.6°C on August 26 and 3.2°C on September 30. These are regional sea-surface anomalies against 1991–2020, not additions that can be applied directly to NASA’s global average (NOAA weekly SST index).

The relative ENSO measure gives a useful restraint on the headline numbers. April–June RONI was 0.5°C and July–September was 1.7°C. For comparison, July–September RONI was 1.8°C in 1997 and 1.5°C in 2015. The event is strong, but not unprecedented in this comparison. RONI removes tropical background warming, which helps separate the ENSO signal from the already warmer ocean backdrop (NOAA RONI).

The subsurface evidence supports persistence. NOAA’s upper-300-m equatorial temperature anomaly for 180°W–100°W rose from 2.26°C in June to 2.96°C in July, 3.23°C in August and 3.38°C in September, against 1981–2010. This is a regional layer-temperature measure, not global ocean heat content or a direct global-temperature forecast. I read it as evidence against a rapid disappearance of Pacific warmth, not as an independent extra warming increment (NOAA subsurface index).

Timing is the key link between those Pacific observations and the annual record. NASA describes a roughly three-month lag between ENSO variability and seasonal global temperatures. Warmth established during summer can therefore affect the remaining months of 2026. Further strengthening toward winter also contributes to 2027, outside this market’s window (NASA discussion of ENSO and global temperature).

September provides the main counter-signal. ERA5’s September anomalies were 0.9263°C in 2023, 0.7237°C in 2024, 0.6582°C in 2025 and 0.8191°C in 2026, against 1991–2020. September 2026 was warmer than the same month in the incumbent record year, but cooler than the exceptional September 2023. It also eased from August 2026’s approximately 0.846°C. October 1–4 averaged 0.6925°C; four days cannot determine the full month, but they argue against assuming uninterrupted acceleration (ERA5 daily series).

Those observations do not translate exactly into NASA temperatures. ERA5 measures global surface-air temperature, while NASA uses a blended air and sea-surface-temperature index (NASA temperature forecasting documentation). A provisional September NASA estimate of 1.37°C is an assumption, not an official observation. At that level, October–December would need to average approximately 1.42°C to reach the rounded tie boundary.

Scientific outlooks also disagree. Berkeley Earth’s August update gives approximately 93% probability of a record in its own dataset and emphasizes that part of El Niño’s warming influence carries into 2027 (Berkeley Earth update). NOAA’s August assessment most strongly favors second-warmest for the year (NOAA annual outlook). Neither is the NASA resolution target. Together they support retaining meaningful uncertainty rather than transferring another agency’s conclusion directly.

What's non-obvious

The question is easier than an outright new record, but harder than “August was hot.” A displayed tie qualifies, lowering the approximate annual hurdle. Yet the early-year deficit means even four months repeating August would narrowly miss. The relevant question is not whether late 2026 is exceptionally warm. It is whether it is warm enough to lift the full-year average across a narrow publication boundary.

The other subtlety is the distinction between Pacific strength and global timing. A very strong El Niño does not guarantee the calendar-year record. The relative index shows that part of the striking conventional Pacific anomaly reflects the warmer background ocean, while the lagged global response spans two years. The strongest case for Yes comes from warmth already established in summer, not simply from projections of a stronger winter peak.

Uncertainties

The main No pathway is an insufficient or delayed global response, not necessarily an El Niño reversal. Cooler land and polar conditions can offset part of the Pacific contribution. September’s easing and the early-October dip keep that pathway credible.

The largest near-term evidence gap is NASA’s unpublished September observation. Its release will remove the need for a cross-product estimate and clarify the remaining-quarter hurdle. Verified numerical forecasts from NASA’s annual temperature forecasting project would provide another direct cross-check; those values remained unverified (NASA forecast project).

Historical comparisons are limited by the small number of strong developing events and by the use of revised data rather than archived first releases. Publication effects also matter: earlier 2026 months and the incumbent record can change before the annual result appears, and NASA’s exact halfway-rounding treatment was not verified. Because this market settles immediately on the first qualifying result, a later scientific revision can leave the ultimate temperature history different from the resolved market outcome.