Why prayer time apps disagree.

The short answer, and why it is incomplete

Prayer time apps disagree because Fajr and Isha have no directly observable clock time. They are defined by how far the sun sits below the horizon — true dawn, and the end of twilight — and different authorities settled on different angles for that. An app is not calculating something wrong; it is applying a different convention.

That is where most explanations stop, and it leaves the useful question unanswered: how much does the choice actually change? We measured it. Every figure below was computed with adhan-js — the same engine behind this site’s own prayer times — over the 2026 calendar year.

The five methods measured

Umm al-Qura is the odd one out and worth noticing: it does not use an angle for Isha at all, but a fixed 90 minutes after Maghrib. That is why it drifts away from the others in a different pattern — its Isha does not care how fast twilight is fading at your latitude.

Finding 1: the disagreement is a function of latitude

The table gives the gap in minutes between the earliest and the latest of the five methods — first on the March equinox, a representative ordinary day, then the widest the Fajr gap ever gets during 2026 and the date it happens.

The ordering is almost perfectly monotonic with latitude, and the reason is geometric rather than doctrinal. Near the equator the sun drops through the twilight angles almost vertically, so the difference between 15° and 19.5° is a few minutes of clock time. The further north you go the shallower that descent becomes, and the same few degrees of angle stretch into hours. In Mecca the choice of method is worth about twenty minutes. In Stockholm at the end of August it is worth over two.

This is the practical takeaway for anyone comparing two apps: near the equator, a disagreement of half an hour means one of them is probably misconfigured. In northern Europe, a disagreement of two hours can be two perfectly correct implementations of two different conventions.

Finding 2: past a point, your method stops being applied

Something stranger happens further north, and it is easy to misread. Above roughly 48° of latitude, adhan — like most implementations — stops trusting the angle in summer and falls back to a night-fraction rule, dividing the time between sunset and sunrise instead. It does this quietly, without changing the method name shown in the interface.

Berlin on the 2026 solstice shows what that does. The four angle-based methods, left to their own definitions, give Fajr times spread across 33 minutes. With the fallback in place, all four return the same instant — an hour and a quarter earlier than any of them actually computed:

So in a summer comparison of northern prayer apps, identical times are not evidence that the apps agree, and are not evidence that the method you selected is being honoured. They are evidence that a substitute rule has taken over and your selection has stopped mattering. Two apps can show the same minute here and still diverge by an hour a fortnight later, when the angle becomes reachable again and each returns to its own definition.

Finding 3: far enough north, there is no answer to compute

At Tromsø, 69.6° N, the sun does not reach the required depression below the horizon at all for 69 days of 2026 — from 18 May to 25 July. There is no angle-based Fajr or Isha during that window, in any method. Not a late one: none.

Every time an app displays for those dates is therefore a substitution — nearest latitude, a fraction of the night, or the times from Mecca. Which substitution is right is a question for scholars, and the honest thing for software to do is to say which one it used rather than present the output as an ordinary calculation. It is also why a prayer app that behaves well in Cairo tells you nothing about how it behaves in Tromsø.

What this means when you pick an app

• Check the method before the app. At your latitude the method may be worth two hours; no amount of interface polish compensates for the wrong convention.
• Follow your local authority. The measurement here says how far apart the conventions are, not which is correct — that is a scholarly matter, and the convention followed by your local mosque is the one to match.
• Distrust agreement in northern summer. Identical times above ~48° in June usually mean a fallback rule is running, not that two calculations concur.
• Ask what happens in the polar window. If you are above roughly 65°, the app’s substitution rule is its prayer time calculation for part of the year.

Reproducing this

Every figure comes from adhan-js with default settings, iterating each day of 2026 and comparing MuslimWorldLeague, NorthAmerica, Egyptian, Karachi and UmmAlQura. Gaps are measured between absolute instants rather than clock readings — a detail that matters, because a northern Isha can fall after midnight and a naïve clock comparison reports a twenty-four hour difference where the real one is forty minutes.

Finding 2 compares the default high-latitude rule against HighLatitudeRule.TwilightAngle, which disables the fallback and returns nothing when the angle is unreachable; the difference between the two is the fallback’s effect. Finding 3 counts the days on which that unfiltered configuration returns no valid time at all.

These are measurements of one widely used implementation and its defaults, not universal constants — another library with different high-latitude defaults will produce different numbers for findings 2 and 3, which is itself part of why apps differ. If your own run disagrees with the tables above, please tell us.

Common questions

Why do two prayer time apps show different times for the same place?

Because they are using different calculation methods, which place Fajr and Isha at different angles of the sun below the horizon. Measured across five common methods, the gap is around 20 minutes near Mecca and grows with latitude: 43 minutes in Istanbul at its widest, and about two hours in London, Stockholm and Reykjavík at the height of summer.

Which prayer time calculation method is correct?

None of them is a computational error — each encodes a different scholarly judgement about how far the sun must be below the horizon for true dawn and the disappearance of twilight. The convention followed locally is the one to use; where methods differ by two hours, that choice matters far more than the app.

Why do apps agree in northern summer when they disagree the rest of the year?

They usually have not agreed — the method has stopped being applied. Above roughly 48° of latitude, implementations fall back to a night-fraction rule when the sun stays too high. In Berlin on the 2026 solstice, five methods whose own angles give Fajr between 00:23 and 00:56 all return 23:08 instead: an identical time produced by the fallback, not by any of them.

What happens at latitudes where the sun never sets?

The angle simply cannot be reached, so no angle-based time exists. In Tromsø the sun fails to reach the required depression for 69 days of 2026, from 18 May to 25 July. Every time shown for those days comes from a substitute rule, and which substitute is chosen is a scholarly question, not a technical one.