El Niño and East Coast Skiing 2026/27: It Takes the Cold, Not the Snow

A very strong El Niño is a temperature event in the East, not a snowfall event. Across seven long-running stations in Vermont and New Hampshire, every one of the three very strong El Niño winters on record lost cold nights and gained thaw days. Snowfall did not move in any consistent direction at all.

That matters more here than it would out West, for two reasons that have nothing to do with the Pacific. Eastern ski terrain sits low enough that the entire mountain lives inside the marginal temperature band, and most of the snow your family actually skis on was made by a machine that needs cold nights to run.

Only the very strongest events show up. Weak and moderate El Niño winters are indistinguishable from ordinary winters in our records. This one is forecast to be neither weak nor moderate.

NOAA's Climate Prediction Center issued its most recent ENSO Diagnostic Discussion on 13 August 2026. It carries an El Niño Advisory, a July Niño-3.4 index of +1.4 °C, and sea surface temperature anomalies above +2.0 °C in the eastern equatorial Pacific.

CPC puts the chance of a very strong event this winter above 90%, and the chance of a historic one during October to December at 69%, meaning stronger than any El Niño back to 1950. That historic threshold is measured on the Relative Oceanic Niño Index rather than the more commonly quoted ONI, which is a distinction most coverage will get wrong.

CPC also attaches a hedge that we are going to repeat in every one of these pieces, because it is the honest frame for everything below: with an event this size, the chances of experiencing impacts consistent with El Niño are larger, but they are not guaranteed. The next discussion lands on 10 September 2026.

The chain is worth stating, because an outcome you cannot explain is a coincidence you have decided to trust.

Warm water pools in the central and eastern equatorial Pacific. The thunderstorm activity that normally sits over Indonesia shifts east with it, which relocates an enormous heat source in the atmosphere. That relocated heat strengthens and extends the subtropical jet east across the southern United States, and it pulls the polar jet north.

Both halves of that matter to New England, and they matter in different ways. The southern jet is the one that drives storms up the eastern seaboard, which is why very strong El Niño winters are associated with large coastal storms in the mid-Atlantic. The northern jet is the one that delivers Arctic air into northern New England, and when it retracts, the cold-air supply that Vermont and New Hampshire depend on is weaker and arrives less often.

So the East gets the storms and loses the cold, which is a fair description of what our stations actually recorded. NOAA's own composite of past El Niño winters puts the northern tier of the country, New England included, in the warm zone.

One technical detail that will separate honest coverage from the rest this autumn. In July the Niño-1+2 region ran at +2.9 °C while Niño-3.4 ran at +1.4 °C. That gradient is the signature of an East-Pacific event, in which the heat source sits further east, the subtropical jet extension is stronger, and the whole response over North America is sharper. If the September or October discussions show that gradient collapsing, the assumption behind this article weakens and we will say so.

For our western pieces we used the automated snowpack network, which measures how much water is sitting on the ground as snow. That is the right question in Utah, where the season is simply whatever fell out of the sky and stayed.

It is close to the wrong question in Vermont. There is no snowpack telemetry in the East anyway, but more importantly, the number that decides an eastern family's week is not how much snow fell. Stowe covers 83% of its terrain with snowmaking. Killington covers 71%. Sunday River covers 95% of its trails. Hunter Mountain covers 100% of its mountain and has called itself the snowmaking capital of the world since it became the first resort anywhere to do it top to bottom.

So we pulled the daily record instead, from the National Oceanic and Atmospheric Administration's Global Historical Climatology Network, and counted the things that actually govern an eastern winter:

• Cold nights, days when the minimum temperature dropped to 23 °F (−5 °C) or below. Snowmaking needs a wet-bulb temperature around 27.5 °F (−2.5 °C), which a night that cold clears comfortably at any humidity.
• Thaw days, days reaching 40 °F (4.4 °C) or above.
• Freeze-thaw days, above freezing by day, well below it by night. The cycle that turns a soft surface into the boilerplate the East is famous for.
• Natural snowfall and peak snow depth, and how many days the ground held a real base of 12 inches (30 cm) or more.

This is the part that most El Niño content skips, and skipping it produces a confident answer that means nothing.

The three very strong El Niño winters are 1982-83, 1997-98 and 2015-16. They are spread across a record in which the Northeast has warmed. So finding that all three were warmer than the long-run average proves nothing whatsoever: any three winters drawn from the back half of the record would tend to beat an average that includes the front half.

So we did not compare them to an average. For every station and every measure, we fitted a line through the ENSO-neutral winters only across the whole record, and scored each El Niño winter against where that line sat in its own year. The comparison is against ordinary winters of the same era, with the trend removed. ENSO-neutral is defined from CPC's own published index table.

Counts are shown as the difference from an ordinary winter. Snowfall and depth are shown as a percentage of one. The rank row asks where each of the three winters finished among a pool of 28, made up of every ENSO-neutral winter in the same records plus these three.

Snowfall does not behave. 1982-83 came in at 70% of an ordinary winter, 2015-16 at 43%, and 1997-98 at 102%. That middle winter finished 16th out of 28 for snowfall, which is another way of saying it was completely unremarkable. Run the odds on all three landing where they did and you get 0.17, which is no result at all.

Temperature behaves. All three winters lost cold nights. All three gained thaw days, two of them finishing 1st and 2nd out of 28. The same test that finds nothing in the snowfall column returns 0.011 for thaw days, 0.017 for cold nights and 0.017 for the depth of snow that actually survived on the ground.

And look at what 1997-98 did, because it is the most instructive winter in the table. A normal amount of snow fell, 102% of an ordinary winter. The mountains still finished the season with only 82% of the usual peak snow depth. The snow arrived and did not stay.

Real talk: this is three winters. Three is a small number, we are not going to pretend otherwise, and no test can rescue a sample that size. What we can say is that the three of them agree with each other about temperature and flatly contradict each other about snowfall, and that is a meaningful difference between two columns measured from the same thermometers on the same days.

This page went up with a sixth column, counting the days each winter held a real base of snow on the ground. It is not there any more, and the reason is worth stating rather than quietly editing out.

Building the same analysis for Canada turned up a station whose snow-depth record reads in the hundreds of centimetres for thirty years and then reads 13, 15, 14, 12. No mountain loses its snow like that; an instrument or an observer changed. Checking for that across every station exposed two mistakes of our own. A season in which a station never reported snow depth was being counted as a season with zero days of base rather than as a season with no measurement. And a season was being thrown away entirely if any one of its measurements was patchy, even for the numbers that did not use it.

Fixing the second of those added a seventh station, Peru in southern Vermont, whose snowfall record goes back to 1951 and was being discarded because its thermometer record is thin.

The base-days figure moved every single time we improved the pipeline. That is what an unreliable measurement looks like, and it depends on the most sparsely reported thing in the whole dataset, so we removed it rather than keep publishing a number we did not trust. Peak snow depth measures the same idea from a better-reported instrument, it is still in the table, and it is significant.

Nothing about the finding changed. Cold nights, thaw days and snowfall are the same as the day this went up. If anything the case is stronger: peak snow depth now comes out at the same 0.017 as the temperature columns, while snowfall still returns nothing.

We had the whole record open, so we ran every El Niño winter since 1950, not just the big three, sorted by how strong the event actually was. This turned out to be the most useful thing we found.

A weak El Niño does nothing in Vermont. Fifteen of them in the record, and the cold-night count comes out slightly above an ordinary winter. A moderate one barely registers. The response only becomes large in the top row, and then it becomes very large: eleven fewer cold nights and a fortnight more thawing.

This is worth knowing because it means most of what you may remember about El Niño winters in the East is not evidence about this one. There have been 22 El Niño winters since 1950 that were not very strong, and as far as our stations are concerned they were just winters.

It also means the usual reassurance, that El Niño years in the Northeast are a mixed bag, is true and beside the point. It is a mixed bag because it is averaged over events that were too small to do anything. CPC currently puts this winter above a 90% chance of landing in the bottom row.

The honest caveat runs the other way too. That bottom row is the same three winters, so the ladder is not independent confirmation. It rules out the possibility that we picked three unlucky winters out of a generally warm El Niño population. It does not turn three into thirty.

The reason a few degrees does more damage here than in Colorado is not meteorology. It is geography and engineering.

There is no altitude to retreat to. Killington's summit, the highest lift-served point in Vermont, is 4,241 feet. Vail's base area is 8,120 feet. Vermont's highest lift-served terrain tops out nearly 4,000 feet below the bottom of Vail. Out West a warm storm rains on the beginner terrain and snows on everything above it, which is miserable but survivable. In the East a warm storm can rain on the whole hill, summit included, because there is no part of the mountain that is reliably cold when the valley is not.

And the snow is manufactured, which turns a temperature problem into a supply problem. This is the part that gets missed. Snowmaking is usually described as insurance against a lack of snow, and it is: Hunter Mountain can open in a winter that delivers nothing from the sky. But insurance against a lack of snow is not insurance against a lack of cold. The guns need a wet bulb near 27.5 °F to run at all. Eleven fewer cold nights is eleven fewer nights of production, and it lands on top of more days of melting.

Which is why 1997-98 matters so much more than its unremarkable snowfall total suggests. A resort that depends on manufactured snow is not really asking whether it will snow. It is asking how many hours it will be cold enough to work, and how much of that work survives until the weekend.

A weather station is not a ski resort. Some of these sit in a valley several hundred feet below the base lodge, and one of them sits far above any lift in the East. We are listing the distance and the elevation so you can see exactly what you are being told.

Hunter and Windham sit six and seven miles from a station at East Jewett, at 1,990 feet, which puts it between Hunter's base and its summit. The problem is that its record only begins in 1986, so it never saw 1982-83 and it does not contain enough ordinary winters to support the same trend correction. Putting it in the main table would have implied a strength of evidence that is not there.

Here is what it recorded, compared to its own median, with that weaker footing stated plainly. In 1997-98 it measured 117% of its normal snowfall and fewer thaw days than usual. In 2015-16 it measured 40% of normal snowfall, a peak depth of 9 inches against a normal 20, and 107 thaw days against a normal 74.

One winter fine, one winter dreadful, no third data point, no trend correction. For the Catskills the truthful answer is that we do not have enough to tell you, and the regional physics is the same as Vermont's only lower and further south, which cuts both ways: closer to the favoured storm track, and warmer to begin with.

Partway through this we found something that looked excellent. Sorting the same data by holiday week, Christmas appeared to be hit hard while Presidents' week looked untouched. Across 18 station-winters it was overwhelming, and it would have made a useful headline: book February, not December.

It was an artefact of how we were counting. Six weather stations inside one small region during the same winter are not eighteen independent observations. They are three weather events measured six times each. When we redid it treating each winter as one observation, which is the only defensible unit when you have three of them, the Christmas effect fell apart. 1997-98 had a perfectly normal Christmas week.

So we are not telling you to book February. We are telling you that we checked, that the check failed, and that a site confidently advising you on which holiday week to pick in an El Niño winter is working from the same data we are.

None of this is a reason to cancel. Eastern resorts have spent sixty years engineering around exactly this problem and they are extremely good at it. But there are choices where this changes the answer.

Snowmaking coverage is worth more than snowfall averages this year. In a normal winter you might reasonably favour Jay Peak's natural snowfall. In a very strong El Niño the resort with the bigger snowmaking system and the deeper cold has more ways to save your week.

Elevation still buys you something, just less than out West. The Mount Mansfield station at 3,950 feet lost 8 cold nights across the three winters. North Conway at 520 feet lost 18. Higher, north-facing terrain held up measurably better.

Early season is the exposed part. Not because Christmas is specially cursed, but because in a warm winter there is no accumulated base to fall back on, and every one of these winters showed fewer days with a real base on the ground. A January or March trip is standing on more snow than a December one.

Book refundable, and check the freezing level rather than the snow total. The forecast number that decides an eastern ski day is the freezing level, not the accumulation.

Pack for wet. The freeze-thaw count went up in every one of these winters. That means firm mornings, soft afternoons, and at least one day where everyone gets damp. Waterproof outer layers rather than water-resistant ones, and a spare pair of gloves per child.

The El Niño answer is close to the opposite in California, it is a rain problem rather than a drought in the Cascades, it is a statistical artefact in the Rockies, and it is not a useful planning input at all across most of Europe. Every region we cover gets an honest answer in our guide to El Niño and the 2026/27 ski season, and if you are weighing the East against a western trip, El Niño and Tahoe covers the strongest positive signal in North America.

If you have already settled on the East, our best East Coast ski resorts for families and Vermont resorts for families guides cover the resort-by-resort choices, and the family ski packing list is built for exactly the wet, firm conditions described above.