The Best Time to See the Northern Lights

The usual answer to this question is "September to March, and go in winter for the long nights". That is not wrong, and it collapses four separate variables into one sentence in a way that loses most of the useful information.

Four things have to line up: a clear sky, enough darkness, no moon in the way, and something happening geomagnetically. They peak at different times of year, and two of them move in opposite directions across the season.

Look at September and December side by side. September has twice the cloud odds and about a fifth of the darkness. December has the most darkness available anywhere and the worst sky in the dataset. Neither is the answer; they are answers to different questions.

October and February are the months nobody recommends and probably should. Both hold a reasonable sky together with six or seven hours of real darkness, which is the combination that actually matters. They are also cheaper than December.

If you have flexibility, two windows in this season stand out, and both for the same set of reasons.

Early to mid September 2026. New moon on the 11th, best cloud odds of the season, and the autumn equinox nearby, which through the Russell-McPherron effect makes geomagnetic storms roughly twice as likely as at the solstices. The cost is darkness: you get a short window each night, and in Tromso a very short one.

Early March 2027. New moon on the 8th, the spring equinox approaching, and Rovaniemi's best month by a wide margin at 50.0%. Darkness is back up to a usable 4.3 hours there. This is the strongest single combination in the season, and it is shoulder-season pricing.

Both are cheaper than December, and both are better bets. The only thing December has is school holidays, which for many people is the deciding factor and there is no arguing with it.

Within a night, the window to be outside is around magnetic midnight, usually between 21:00 and 01:00 local time. That is when your location rotates under the most active part of the auroral oval, which stays fixed relative to the sun while the earth turns beneath it.

It is a real effect and not a strong enough one to be the only thing you plan around. Displays happen outside that window regularly. But a tour that returns to the hotel at 21:30 is selling you the beginning of the useful period rather than the middle of it, and that is worth noticing in a listing.

More than one. This is the single most useful piece of planning advice on the site, and it follows directly from the cloud figures.

At Tromso's December rate of about one usable night in six, a single booking is a poor bet no matter how the listing is worded. Three nights changes the arithmetic completely. Two cheap nights beat one expensive guaranteed one, and the reason is arithmetic rather than opinion.

Most people cannot choose freely. School terms, work, and flight prices decide the month, and the only remaining variable is where to go. On that reading the table above answers a different and more useful question.

Reykjavik appears three times in a row through the middle of the season, and that is not a preference, it is what the data says. November to January is when the Atlantic weather that plagues Tromso is at its worst and Iceland's consistency is worth the most.

The counter-argument, and it is legitimate: Iceland sits further from the auroral oval, so a quiet night shows less there than in Tromso. If your priority is the strongest possible display and you have several nights to gamble with, Tromso still has the better ceiling. If your priority is seeing something at all, on a fixed date, the table above is the honest answer.

Deciding when to go and deciding when to buy are separate problems, and they pull in opposite directions.

Flights and accommodation reward booking early, especially for anything touching the Christmas fortnight, when prices in Tromso and Rovaniemi rise steeply and availability disappears.

The tour itself rewards booking late. Once you are on the ground you can see a two or three day forecast, which is the only forecast horizon that means anything, and you can pick your night. The cost of waiting is that the good small operators, the ones running long chases with eight people, sell out first.

The compromise most experienced visitors settle on: book flights and hotel months ahead, book one tour in advance for insurance, and keep a second evening free to book on the ground once you can see the weather. That gives you two attempts, one of which is chosen with information.

It is also worth knowing that free cancellation windows are typically 24 hours, which is not long enough to be a weather strategy on its own. At 24 hours the cloud forecast is decent but not decisive, and you are usually choosing between going out under a doubtful sky and losing the slot.

• Cloud figures: ERA5 reanalysis (ECMWF, Copernicus Climate Change Service) via the Open-Meteo historical archive, eleven seasons September 2015 to April 2025. Dataset at the Copernicus Climate Data Store. Full method on our weather page.
• Darkness hours and moon phases: calculated from standard astronomical algorithms; method on our moon and darkness page.
• Equinox effect: Russell, C. T. and McPherron, R. L. (1973), Journal of Geophysical Research, doi 10.1029/JA078i001p00092. Geomagnetic storms are roughly twice as frequent near the equinoxes as near the solstices.
• Solar cycle context: NOAA solar cycle progression and NASA.
• Forecasts for the night itself: MET Norway and the Icelandic Met Office.