Best Month to See the Northern Lights in Iceland: A study of 95 years

We built this study to settle which month actually gives you the best chance of northern lights in Iceland, using 95 years of geomagnetic records rather than guesswork. It covers which months are the most magnetically active, why activity peaks twice a year, how often severe storms hit in each month, how all of that lines up with the hours of darkness you really get, and what our companion cloud-cover study says about actually seeing any of it.

Also read: which month you go matters less than where you stand. Our companion study Iceland’s clearest skies: the best regions for northern lights analysed 50,087 cloud observations from 261 Icelandic weather stations to rank the best places in the country to actually see the aurora. The regional spread is more than six times wider than the monthly one.

Every aurora forecast you have ever read rests on one number: the Kp index. It runs from 0 to 9 and describes how disturbed Earth's magnetic field is over a three-hour window. It is what the guides on northern lights tours are watching when they decide which way to drive, and it is why northern lights hotels away from the city glow are worth the extra distance. Most advice on which month to book, though, is written from habit rather than from the record.

This study analyses 276,584 three-hourly Kp readings covering 95 years, from 1 January 1932 to 27 August 2026, published by the GFZ Helmholtz Centre for Geosciences in Potsdam. It is the longest unbroken space-weather record that exists, and the only one long enough to answer a question about months.

Which month is the most active

March is the most geomagnetically active month over Iceland, averaging Kp 2.39 across 95 years. September comes second at 2.36 and April third at 2.36. At the other end, December is the quietest month of the entire year at 1.94.

Kp is a compressed scale, so the gap between 2.39 and 1.94 looks smaller than it is. Converted to ap, the linear index published alongside Kp, March averages 31.8 nT of magnetic disturbance against December's 21.2 nT. March runs 50% more disturbed than December, year in and year out.

March, April and September sit clearly above every other month, and their confidence intervals overlap each other, so we can name the leading group but not rank inside it. February and October form a second tier. Then the record drops away to the solstice months.

April is the awkward one. It ranks near the top for magnetic activity and is close to useless for actually seeing anything, because by mid-April Reykjavik has barely five hours a night dark enough. We come back to that below.

Why activity peaks twice a year

The double hump is not noise and it has nothing to do with Icelandic weather. It is geometry, and astronomers have seen it in magnetic records since 1856.

Energy only crosses into Earth's magnetosphere when our magnetic field and the Sun's point in opposite directions and reconnect. Near the equinoxes, the tilt of Earth against the spiral of the interplanetary magnetic field makes that southward component larger on average, reconnection happens more readily, and Kp runs higher. Physicists call it the Russell-McPherron effect, after the 1973 paper that set out the mechanism.

It is not the only thing going on. Cliver, Kamide and Ling fitted three competing mechanisms to 130 years of magnetic data and found the equinoctial hypothesis carried the most weight, ahead of Russell-McPherron and the older axial hypothesis. Lockwood and colleagues went further in 2020 and declined to crown a winner, concluding the seasonal signal is a mixture of all three.

The smoothed peaks lag the equinoxes by roughly five days, near 26 March and 24 September. The exact date shifts by up to a week depending on how you smooth the data, so it is not worth quoting more precisely than that. What matters for a trip is the shape: activity stays elevated for about six weeks either side of each equinox, and that is a window you can plan around.

How often the big storms happen

Monthly averages flatten the thing travellers actually care about, which is the rare night when the whole sky goes active. So we counted the severe intervals instead: every three-hour window in 95 years that reached Kp 7 or higher.

September recorded 221 of them and March recorded 214. December recorded 23. That is a 9.6-fold gap between the strongest and weakest months of the aurora season, measured on exactly the displays people cross an ocean for. September and March between them account for close to a third of every severe interval on record. December accounts for 2%.

One honest caveat: 11 of the 13 observatories that produce Kp sit in the northern hemisphere, so some part of the December deficit may reflect winter conditions at the measuring stations rather than the magnetosphere itself. The pattern is rock solid in the data. The cause is still argued over.

Darkness is the other half of the equation

Kp on its own cannot tell you when to come, because you need a dark sky as well as an active one, and in Iceland those two peak at opposite ends of the season. So we combined them. For every one of the 276,584 readings we computed the sun's elevation at Reykjavik and kept the portion of each interval where it sat at least 10 degrees below the horizon, the point where aurora becomes visible.

December offers 15.5 hours of usable darkness a night and November 14.1, more than any other month. But December converts only 0.39 of those hours into Kp 5 or better. October converts 0.77 hours, twice as many, out of 11.4 hours of darkness. March manages 0.68 out of 9.3, February 0.65 out of 12.5, and September 0.61 out of 8.1.

June and July have no astronomical darkness at all, so no amount of magnetic activity is visible. If you are visiting then, the midnight sun is the trade you are making.

Put plainly: the deep-winter months give you more hours to wait in, and the equinox months give you more reason to be outside.

There is a third factor, and we have measured that one too. Our cloud-cover study analysed 50,087 cloud observations from 261 Icelandic weather stations going back to 1937, and it complicates the picture above. September and October, the two months the Kp record likes most, are also the two cloudiest months of the aurora season, averaging 5.96 and 5.97 oktas of cloud across the country. November is the clearest at 5.79.

Before rewriting a trip around that, look at the size of the effect. Across the season the monthly range is 0.19 oktas, about two percentage points of sky. The regional range is 1.22 oktas, more than six times larger. Djupivogur in its cloudiest month is still clearer than Akureyri in its clearest. Which month you pick moves the odds a little. Where you stand moves them a lot.

Does it hold up across solar cycles?

A seasonal pattern in space weather is easy to fake by accident. Solar cycles run roughly 11 years and vary enormously in strength, from a mean Kp of 2.50 in cycle 21 down to 1.57 in cycle 24, so one strong March in one strong cycle can drag a 95-year average around.

We cut the record into nine solar cycles and ranked the months inside each one separately. March lands in the top three in eight of the nine. September also lands in the top three in eight of the nine. April manages seven. December is the weakest month of its own cycle in seven of the nine, and never once reaches the top three.

That is about as stable as a seasonal signal gets in this field. It also means the finding is not a product of the current solar cycle and will not expire with it.

So which month should you visit?

We recommend late February to late March, or October and November. Iceland in March is the only month of the year that wins on all three measures at once: it is the most magnetically active month, it still has 9.3 hours of darkness a night, and its skies are clearer than the season average. Iceland in October leads on the single most useful number, active dark hours, and Iceland in November has the clearest skies of the whole season with 14 hours of darkness to use them in. February sits just behind March on every count.

September is the month our two studies disagree about, and it is worth being straight about that. On magnetic activity Iceland in September is excellent, effectively tied with March. On cloud it is the second-worst month of the season, and inside the southeast corridor it is the worst of all at 5.70 oktas. September is still a good month to come. It is just the month where the choice of base matters most, which is what the clear-skies study is for.

There is a second reason to prefer the shoulder months, and it is the one most aurora advice gets backwards. Iceland sits at roughly 64 degrees corrected geomagnetic latitude, which puts the country under the auroral oval rather than south of it. At Kp 2 or 3 the aurora is already overhead here. Chasing the highest possible Kp is a mid-latitude strategy: at Kp 5 and above the oval expands southward and can leave Iceland on its poleward edge, with the display sitting to the south instead of arcing overhead. Higher is not automatically better at this latitude, which is worth knowing before you read too much into a big forecast number.

Which leaves the practical trade. Iceland in December has the longest nights and the busiest tour calendar, and if you are here for Christmas lights and ice caves with aurora as a bonus, it works fine. If aurora is the whole reason for the trip, come in the shoulder season instead. You will get a more active sky, more mobile weather between fronts, cheaper accommodation, quieter roads, and, in late September and October, autumn colour in the highlands as well. Our guide to the best time to visit Iceland covers the rest of the seasonal trade-offs.

Where to stay and how to chase it

Kp tells you what the magnetosphere is doing, not whether you will see it through the sky above you. Cloud is the variable that actually decides most Icelandic nights, and it is the one a traveller can plan around, so we measured it separately.

Our analysis of Iceland's clearest skies ranked 261 weather stations across the aurora season, and the answer is not where most guides point. The clearest skies run along a corridor in the southeast, from Kirkjubaejarklaustur through Jokulsarlon and Hofn to Djupivogur. Djupivogur averages 5.42 oktas of cloud across the season, the lowest in the country, and recorded a mostly clear month in 36% of winters. Akureyri, which most aurora advice recommends, averages 6.16 and managed it in 3%. Myvatn at 5.65 is the strongest northern base. One caveat from that study worth repeating: clearest is not driest, and the southeast is the wettest region in the country.

The practical version is this. Base yourself between Kirkjubaejarklaustur and Djupivogur if clear sky is your priority. Book a hunt that drives rather than a fixed viewing spot, which is what most of our northern lights tours are built to do, or take a rental car and check the cloud map each afternoon so you can move to the gap. And sleep somewhere with dark skies and a wake-up call: our northern lights hotels are chosen for exactly that, and if you want the shortlist with photos and locations, see our roundup of the best northern lights hotels in Iceland.

For the practical side of a hunt, read how to find the northern lights in Iceland, our overview of weather and northern lights in Iceland, and, if you want to bring pictures home, how to photograph the northern lights. If you are new to the phenomenon itself, start with the northern lights in Iceland.

Summary

• Most active month: March, at a mean Kp of 2.39, with September and April a hair behind.

• Quietest month: December, at 1.94, which is also Iceland's darkest month.

• Best months to book: late February to late March, or October and November, where activity, darkness and clear sky overlap best. March wins on all three.

• Severe storms: September logged 221 intervals at Kp 7 or above across 95 years. December logged 23.

• Month to skip if aurora is the point: April, which is active but has run out of darkness, and June and July, which have none.

• Do not chase high Kp: at Iceland's latitude, Kp 2 to 3 already puts the aurora overhead.

• Cloudiest months of the season: October at 5.97 oktas and September at 5.96. November is the clearest at 5.79.

• Where beats when: the regional cloud spread is more than six times the monthly spread, so base yourself in the southeast corridor.

How we did this study

The source is the Kp_ap_since_1932 series published by the GFZ Helmholtz Centre for Geosciences, Geomagnetic Observatory Niemegk, under a CC BY 4.0 licence. We retrieved it on 28 August 2026. It contains 276,584 three-hour Kp values from 1 January 1932 to 27 August 2026 with no gaps: every single interval carries a value. The final 216 intervals, from 1 August 2026, are flagged preliminary rather than definitive.

We computed monthly means over all intervals, and again over the night intervals of 21:00 to 03:00 UTC, since Iceland keeps UTC year-round. Confidence intervals come from a 2,000-sample bootstrap of the monthly night-hour distributions. Solar elevation at Reykjavik (64.1466 N, 21.9426 W) was sampled every five minutes, giving each three-hour interval a dark fraction rather than a yes or no flag, with dark meaning the sun 10 degrees or more below the horizon.

Every monthly figure was recomputed independently in a second language before publication, and both passes agree to three decimal places. Where we quote a magnitude as a percentage, we use ap rather than Kp, because Kp is quasi-logarithmic and its differences cannot be read as ratios.

Cloud-cover figures quoted here come from our companion study, Iceland's clearest skies: 50,087 monthly cloud observations from 261 Icelandic Met Office stations, 1937 to 2026, with the month-by-month figures covering 1961 to 2010. Oktas are eighths of the sky, so 5.96 oktas is about 75% of the sky covered.

Two limits are worth stating plainly. Kp is a planetary index built from 13 subauroral observatories between 44 and 60 degrees geomagnetic latitude, weighted towards Europe and North America, so it is not an Iceland measurement. And it says nothing at all about cloud, which is the binding constraint on most Icelandic nights and the reason we measured that separately. Averages describe a season, not a night: a disappointing March and a spectacular December both happen every few years.