Easter Date and the Computus That Sets It

Easter Sunday 2026 falls on April 5

Easter Sunday falls on April 5, 2026 in the Western churches and on April 12, 2026 in the Orthodox churches. That is day 95 of the year, with 270 days left in 2026. The gap between the two dates is one week, which is one of the three gaps that occur: zero days, seven days, or thirty-five days.

The date is tied to the moon rather than fixed to a calendar day. Easter is the first Sunday after the paschal full moon, which is the first ecclesiastical full moon on or after March 21. The moon returns to the same phase about 11 days earlier each year, and the weekday of March 21 moves as well, so the result shifts around the calendar by as much as a month from one year to the next.

Every movable observance in the liturgical year hangs off this date. Ash Wednesday sits 46 days before it, Good Friday 2 days before, Ascension 39 days after, and Pentecost 49 days after. Set Easter and the rest of the calendar follows.

Those seven dates are enough to plan a spring schedule. The calculator above returns the same list for any year from 1583 onward, which is the first full year of the Gregorian calendar.

The rule, and why the window is 35 days

The rule that fixes the date was settled at the Council of Nicaea in 325. Every church was to keep Easter on the same Sunday, and the Sunday chosen was the one after the full moon that follows the spring equinox. The wording leaves two things open, and the medieval computists closed both of them.

The equinox is fixed at March 21 even though the astronomical equinox drifts about a day every 300 years against the calendar, as the US Naval Observatory note on the date of Easter sets out. The moon is the ecclesiastical moon, a value computed from a 19-year cycle rather than an observed phase. Those two choices turn a set of observations into arithmetic that a scribe in any monastery could repeat without an instrument.

The result is a 35-day window. The earliest Easter is March 22, which needs a full moon on March 21 that falls on a Saturday. The latest is April 25, which needs the paschal full moon on April 18 to fall on a Sunday. Both ends are rare. The most recent March 22 Easter was 1818 and the next is 2285. The most recent April 25 Easter was 1943 and the next is 2038.

The earliest disputes were not about arithmetic at all. Quartodeciman congregations in Anatolia kept the feast on the fourteenth day of Nisan whatever weekday that was, following the practice that Polycarp of Smyrna traced to the apostle John. Victor of Rome excommunicated them for it around 190, and the quarrel rumbled for another century and a half until Nicaea settled on a Sunday for everyone.

The 35-day span is why Easter can land on any of 35 calendar dates but never on March 21 or April 26. A date outside that range is a sign that someone has confused Easter with a solar feast such as Christmas.

The Meeus/Jones/Butcher computus in nine steps

Jean Meeus published this version of the computus in Astronomical Algorithms in 1991, adapting a method that Samuel Butcher set out in 1876. It is the standard algorithm for the Gregorian calendar because it needs only integer arithmetic and no tables. Every division is a floor division and every mod is a non-negative remainder.

The nine steps run in order: reduce the year to a position in the lunar cycle, split the century from the year within it, apply the solar correction, apply the lunar correction, then convert the final number into a month and a day. The table below shows each step and the value it produces for 2026.

Step 5 is the one that drifts. The expression carries the epact, the notional age of the moon on January 1. The terms d and g are the solar and lunar corrections that keep the 19-year cycle in step with the tropical year. Drop them and the computed date walks out of the March 22 to April 25 window within a few centuries, which is exactly the error the Gregorian reform was built to fix.

Worked example: the arithmetic for 2026 and 2027

Running the steps by hand for 2026 shows where each number comes from. Work in integers and keep the remainders.

Start with a. Nineteen times 106 is 2014, so 2026 mod 19 leaves 12. The century split gives b = 20 and c = 26. Halving the century gives d = 5 with e = 0, and the correction terms give f = (20 + 8) / 25 = 1 and g = (20 - 1 + 1) / 3 = 6.

Now h: 19 times 12 is 228, and 228 + 20 - 5 - 6 + 15 = 252. Since 30 times 8 is 240, the remainder is 12, so h = 12. The century leap count gives i = 6 and k = 2, which puts L = (32 + 0 + 12 - 12 - 2) mod 7 = 30 mod 7 = 2.

The last step is arithmetic on one number: h + L - 7m + 114 = 12 + 2 - 0 + 114 = 128. Dividing 128 by 31 gives 4 with a remainder of 4, so the month is April and the day is 4 + 1 = 5. Easter 2026 is April 5.

Two of those columns carry more meaning than their labels suggest. The paschal full moon is March 21 plus h, so h is a count of days rather than a calendar date. Easter then falls L + 1 days after that full moon, which is why 2025 has L = 6 and still lands seven days after a full moon that already fell on a Sunday.

A second run for 2027 checks the method. Here a = 13, so 19 times 13 makes 247, and 247 + 20 - 5 - 6 + 15 = 271. Since 271 mod 30 = 1, the value of h is 1. With L = (32 + 0 + 12 - 1 - 3) mod 7 = 5, the combined value is 1 + 5 + 114 = 120. That is 3 with a remainder of 27, so the month is March and the day is 28. Easter 2027 is March 28, eight days earlier than 2026.

Western and Orthodox Easter use different calendars

The Orthodox churches keep the same rule but run it on the Julian calendar, then convert the answer to the Gregorian calendar for civil use. In the twentieth and twenty-first centuries the Julian calendar runs 13 days behind, so the same nominal date lands 13 days later on a modern wall calendar.

There is a second difference. The Orthodox rule requires that Easter fall after the Jewish Passover, so when the two would collide the date moves forward by a week. That is why the gap between the two Easters is 0, 7, or 35 days rather than a fixed 13 days.

The conversion uses the century: the offset is (century - century div 4 - 2) days, which gives 13 for 1900 through 2099 and 14 from 2100. The widget above applies that offset to the Julian result, so the Orthodox figure it reports is already a Gregorian date.

Two rival tables circulated through the first millennium. Rome ran an 84-year cycle whose Easter drifted against the moon, while Alexandria ran the 19-year cycle with the saltus lunae, a one-day jump inserted at the end of each cycle to keep the reckoning in step. The Alexandrian tables won. Dionysius translated them into Latin in 525, Bede popularised them in Northumbria two centuries later, and the paschal letters that bishops circulated each year were read from that lineage until Clavius replaced the tables with arithmetic.

The Gregorian side has its own history. Aloysius Lilius drew up the reform and Christopher Clavius defended it, and Pope Gregory XIII published it in the bull Inter gravissimas in 1582. That October, Thursday the 4th was followed by Friday the 15th, ten days vanishing from the civil calendar in one step. The computus changed with it: the epact tables were rebuilt and the two correction terms in step 5 of the algorithm were introduced for the first time.

Several Orthodox churches adopted the Revised Julian calendar at a congress in Constantinople in 1923, which aligns their fixed feasts with the Gregorian calendar until 2800. Easter was deliberately excluded from that change, so the Greek, Russian, Serbian and Antiochian churches still compute the feast by the Julian rule even where they keep Christmas on December 25 by the modern calendar. That split is why a Greek parish may hold Nativity on the same day as a Latin one and still keep Easter a week or five weeks later.

Four of the twelve years in that table coincide, which is close to the long-run rate. When a year has both dates on the same Sunday, the two calendars have absorbed the 13-day offset inside the lunar correction rather than outside it.

The movable feasts that follow Easter

Once Easter is fixed, a dozen other dates follow by simple addition. None of them needs its own calculation, which is why a table of offsets is enough to build a whole liturgical calendar.

Lent runs 40 fasting days that begin on Ash Wednesday, and the 46 days between Ash Wednesday and Easter include the six Sundays that are not counted as fast days. That is why the offset is 46 and not 40, and it is the detail most people get wrong when they count backwards from Easter.

Ascension is 39 days after Easter because it falls on the fortieth day when Easter itself is counted as day one. Pentecost is 49 days after, seven weeks later. Trinity Sunday is the Sunday after Pentecost, and Corpus Christi is the Thursday after Trinity, which puts it 60 days after Easter.

Because every offset is a whole number of days, the weekday of each observance is locked to the weekday of Easter. Easter is always a Sunday, so Ash Wednesday is always a Wednesday and Ascension is always a Thursday. A calendar that shows Ash Wednesday on any other weekday has a bug.

The 5,700,000 year cycle and the 19-year trap

The computus repeats exactly every 5,700,000 years. That number is the least common multiple of the 400-year Gregorian solar cycle and the lunar corrections, so the pattern of Easter dates is finite and fully known. The calendar itself is a civil standard rather than an astronomical measurement, which is why the NIST time and frequency division treats leap rules as a matter of definition. A table of dates needs no end date.

The obvious shortcut is the 19-year Metonic cycle, and it fails for this calendar. Nineteen Julian years hold 235 lunations to within a couple of hours, so the moon phase repeats, but the weekday does not. Nineteen Gregorian years hold 6,939 or 6,940 days depending on where the leap days fall, and 6,939 mod 7 is 2 while 6,940 mod 7 is 3. The date therefore slips by two or three days each cycle. Between 2024 and 2099 not one year repeats its Easter date exactly 19 years later.

Over the full cycle the 35 possible dates are far from equally likely. April 19 appears in 220,400 of the 5,700,000 years, which is 3.87 percent and the highest share. March 22 appears in 27,550 years, 0.48 percent, and is the rarest. April 25, the latest date the rule allows, appears in 42,000 years, 0.74 percent, the second rarest. The distribution is uneven because the lunar correction and the weekday term interact rather than vary independently.

The tables that made this work predate the algorithm by a thousand years. Dionysius Exiguus produced his Easter tables in Rome around 525 and, while he was about it, fixed the numbering of years from the Incarnation that the West still uses. Bede extended the method in De temporum ratione in 725, and the Venerable one's tables were copied across Europe for centuries before anyone wrote the rule as a formula.

Leap years enter through two terms. The value i = c div 4 counts the leap days inside the century, and a century year that is not divisible by 400 contributes none. That is why 1900 and 2100 shift the pattern while 2000 does not, and why a table built for one century cannot be reused in the next without the correction.

The moon itself does not cooperate with the calendar. Real full moons drift against the computed ones by up to a day and a half, and the NASA phase tables show the observed syzygies rather than the ecclesiastical ones. That gap is deliberate: the computus needs a rule that every church can apply without an observatory, not a rule that tracks the sky exactly.

Planning around a date that moves

A date that can sit anywhere in a 35-day window is a scheduling problem. Retailers plan chocolate and egg inventory around it, and a late Easter pushes the selling window into April while an early one lands it in March. Two Easters four weeks apart in consecutive years means two different first quarters for the same store.

School spring breaks are often tied to Easter and shift with it. Airlines price the surrounding weekends differently. Payroll and accounting teams care for a different reason: Easter is not a federal holiday in the United States, so the business effect comes from Good Friday closures and from the school calendar rather than from a statutory day off.

Year over year the jump is smaller than the full window but still large. Across 2024 to 2035 the date moves forward by as much as 20 days (2027 to 2028, and again from 2029 to 2030) and back by as much as 16 (2028 to 2029), with most steps between 8 and 15 days. A planner who assumes a fixed week number will be wrong most years.

The observance is not only ecclesiastical. Easter Monday is a public holiday in England, Wales and much of the European Union, and Germany's school holidays follow it as the Osterferien in each federal state. Spain runs processions through Semana Santa, the week that begins on Palm Sunday, and hotels along the Mediterranean coast price that week as a peak season. In the United States none of this is statutory, but the egg hunt, the bunny and the chocolate aisle arrive on a schedule that retailers set months in advance from the same computus.

Two habits make the date easier to live with. Store it as a computed value rather than a constant, so next year's figure is never a guess. And when a contract, a promotion or a shipping promise says "Easter week", write the actual dates into the document, because the phrase means a different week every year.