mariuscomper.uk RO

Vrancea, 18 November 2026, 02:48

A thousand years of Vrancea earthquakes, and how much is known about each

On 18 November 2026 at 02:48 Romanian time, Vrancea moved again: magnitude 5.3, 130 km down. It is the newest mark on the drawing below, under the threshold of a record that starts in the year 984. The further back you go, the less certain the lines become.

13 of 37

Of the 37 magnitude-7 quakes in the Romanian catalogue, 4 were measured by seismographs and 9 have felt intensities assigned to places from written accounts. Another 13 are dropped, moved or not covered by the European catalogue.

Each row below is a century, each point a quake in ROMPLUS, the Romanian earthquake catalogue (after October 2022, in the USGS catalogue); the higher and larger, the bigger the magnitude. The shape and colour show what evidence there is. Lower the magnitude threshold and watch the years after 1900 fill up far more than the rest.

The record

7.0

What each mark means (the number shows how many at the chosen threshold)

Quakes of 6.5 or more, as a table (all of them are in the CSV file)
DateMwWhat is known
the year 9847.1unchecked
12 May 10226.5missing
15 August 10387.3missing
the year 10917.1missing
12 February 11077.1copied
8 August 11267.1copied
April 11707.3missing
13 February 11967.5copied
10 May 12307.3from accounts
7 February 12587.1missing
the year 13277.3copied
10 October 14467.5copied
29 August 14717.5from accounts
29 August 14737.3missing
24 November 15167.5copied
9 June 15236.5moved
the year 15437.1moved
19 July 15457.1moved
21 August 15526.5moved
10 May 15717.1moved
April 15786.5copied
30 April 15907.3moved
21 April 15957.1copied
22 November 15986.5copied
3 May 16046.8copied
24 December 16057.1moved
13 January 16066.8moved
8 November 16207.5moved
1 February 16377.1copied
19 April 16506.5copied
9 August 16797.5missing
19 August 16817.1copied
12 June 17017.1copied
11 October 17116.5copied
11 June 17387.7from accounts
5 April 17407.3copied
18 January 17786.5copied
18 January 17876.5missing
6 April 17907.1from accounts
26 October 18027.9from accounts
5 March 18126.5copied
10 February 18216.6copied
17 November 18216.5from accounts
26 November 18297.3from accounts
21 April 18356.5from accounts
23 January 18387.5from accounts
the year 18486.5copied
28 October 18546.5copied
16 October 18626.5copied
13 November 18686.8copied
27 November 18686.5copied
25 December 18806.8copied
19 August 18886.5copied
17 August 18937.1from accounts
10 September 18936.5from accounts
4 March 18946.5copied
31 August 18947.1from accounts
11 March 18966.6copied
6 February 19046.6weak
6 October 19087.1measured
29 March 19346.6measured
22 October 19406.5measured
10 November 19407.7measured
7 September 19456.8measured
9 December 19456.5measured
4 March 19777.4measured
30 August 19867.1measured
30 May 19906.9measured

#What a thousand-year catalogue is

Vrancea, in the bend of the Carpathians, is one of Europe's main sources of large deep earthquakes: they start 60 to 200 km down, inside a slab of old plate sinking almost vertically, and shake an area from Bucharest to Moldova and beyond.

ROMPLUS, kept by INFP, Romania's National Institute for Earth Physics, lists Romanian earthquakes back to 984. Its magnitude is Mw, moment magnitude, the modern scale based on the seismic moment, the size of the rupture. For a quake from the 1500s nobody measured anything: the magnitude is worked back from how strongly the quake was felt, where it was felt, and from what survived in writing. Its entries for 984–1979 come from Constantinescu and Mârza's 1980 catalogue, converted to Mw with relations published by Oncescu and colleagues in 1999.

Each entry has a quality grade from A (best) to D. Before 1906 every entry is D, so the grade cannot separate the old quakes from each other. Of the 37 magnitude-7 quakes, 33 are D. This page grades them another way: by what else, outside the catalogue, still supports them.

#How to read the record

The glowing red point is a measured quake; the dimmer one, from the 20th century, has weak parameters. The gold point with a crown of small dots has intensities by place. The hollow ring is a catalogue value that nothing else supports. The struck point was reclassified by the European catalogue, and the pale cross marks a quake missing from it. Quakes of magnitude 6 or more also have a vertical line up to the point.

All but two of the entries before 1900 sit on the same spot, the catalogue's default epicentre (45.7° N, 26.6° E), at round depths of 100–150 km. The drawing places them in time only, never on a map.

At magnitude 7, four entries give only the year (984, 1091, 1327 and 1543): they appear as a bar across the whole year, not as a point on a day.

Why only the disagreements count: for the old Vrancea quakes, the European catalogue EPICA took its values from the Romanian catalogue published in 1999, the common root of both. When it keeps them, it does not confirm them. When it moves or drops them, it says something.

#Lower the threshold: mostly the age of instruments fills up

At the 6.5 threshold the record has 58 entries before 1900 and 10 from 1900 on. At 4 it has 91 and 807. The old part grows 1.6 times, the new part 81 times. Before 1500, any threshold of 6 or lower gives the same 15 entries.

Few small old quakes appear in the catalogue; seismographs can also record quakes that nobody feels. The instrumental part is not complete from the start either: the European SHEEC catalogue estimates that the record of deep Vrancea quakes of magnitude 5.8 or more is complete only after 1900.

The open version of the Romanian catalogue stops on 31 October 2022. For the years after, the record uses the USGS catalogue: 22 quakes with a USGS magnitude of at least 4, 20 of type mb and two of type mww; mb magnitudes are not identical to Mw. The quake of 18 November 2026 is drawn on its own, in blue.

#The Middle Ages: chronicles from Kiev, Novgorod and Moscow

The catalogue has 13 magnitude-7 quakes before 1500. Two studies went looking for their written sources: Tatevossian and Albini (2010), and Pierre and Dominique Alexandre (2012, 2018). Tatevossian and Albini found the sources cited come almost exclusively from Russian chronicles.

Seven of them (1091, 1107, 1126, 1196, 1230, 1327 and 1446) appear in the Alexandres' 2018 table of accepted quakes, but their accounts come only from Russian chronicles (Kiev, Pereiaslav, Novgorod, Moscow). The quake was felt there; that it came from Vrancea is the seismologists' hypothesis, because no source names a place in the region.

Only one, on 29 August 1471, has witnesses on what is now Romanian soil: Brașov, Sibiu, a Moldavian village where the ground split. The study's table has 15 entries.

The other five (984, 1038, 1170, 1258 and 1473) are not in the 2018 table. For three, Tatevossian and Albini found no source, or a doubtful one, and 1473 is a copy of the 1471 quake.

The dates differ too. The studies of the sources put the 1107 quake on 5 February, not the 12th; 1126 on 1 August, not the 8th; 1196 on 12 March, not 13 February. The notes give both: the catalogue's date and the study's.

#1802: the best documented of the old quakes

Bucharest long remembered the "great earthquake" of 26 October 1802, which badly damaged the Colțea tower. It has the richest intensity map of any quake before seismographs: 81 places, read by Tatevossian and Mokrushina (1998) from accounts of the time, with intensity 8–9 at Brașov and Bucharest.

It is also the largest quake in the catalogue, magnitude 7.9. But its magnitude is a catalogue estimate, not a measurement. In all, the 9 big quakes with intensities by place have 240 observation points.

#What the record cannot tell

The record does not show quieter and busier centuries. The 1800s have 19 quakes of 6.5 or more, more than any other century. The difference lies mainly in how much was written down and kept, and the record does not allow centuries to be compared directly.

Seismologists give different thresholds for when the record becomes complete. The European SHEEC catalogue treats Vrancea quakes of magnitude 6.8 and above as complete only from 1750. INFP researchers (Radulian and colleagues, 2023) treat those above 7 as complete from about 1580–1600. Before those dates, the count is what survived in writing.

For the same reasons, the record does not say when the next one comes. For that, Vrancea keeps no clock analyses the gaps between quakes on the same data: the same 13 magnitude-7 quakes since 1700.

#Your year

Type a year (yours, a grandparent's) and the record shows the big quakes of that lifetime and the one before, with what is known about each.

#Questions

How many big earthquakes has Vrancea had in the last thousand years?

The Romanian ROMPLUS catalogue lists 68 quakes of magnitude 6.5 or more in the Vrancea deep-quake zone from 984 to today, 37 of them magnitude 7 or more. All of Romania has 72 entries of 6.5 or more; not all are from Vrancea.

Are these numbers certain?

Not equally. Only 13 of the 37 magnitude-7 quakes were measured by instruments or have felt intensities assigned to places from written accounts. The rest are catalogue estimates; the European catalogue drops, moves or does not cover 13 of them.

Was the quake of 18 November 2026 a big one?

Not in this record's sense. It was magnitude 5.3, below the 6.5 threshold. The last Vrancea quake of 6.5 or more was on 30 May 1990; since then the largest was 6.0, on 27 October 2004.

What was the largest Vrancea earthquake?

In the catalogue, 26 October 1802, with an estimated magnitude of 7.9. The largest measured by instruments was 10 November 1940, magnitude 7.7.

Does this mean a big one is coming?

This page does not answer that question or hint at it. The gaps between quakes are analysed on Vrancea keeps no clock.

On the same subject

#Method

The quakes come from INFP's ROMPLUS catalogue, version 2 as published on Mendeley Data (2022, CC BY 4.0 licence), compared row by row with INFP's current file (to February 2026): the 68 Vrancea entries of 6.5 or more are identical. The zone: 45.2–46.1° N, 25.9–27.0° E, at least 50 km deep, the same as on Vrancea keeps no clock. Left out: three shallow quakes (1550, April 1571, August 1590) and 25 May 1912 (Mw 6.7, 90 km), which lies east of the zone at 27.2° E.

Each quake of 1000–1899 was looked up in the European EPICA v1.1 catalogue (INGV, 2021, CC BY 4.0) by day, month and year. Two entries appear there on another day and were matched by hand: 19 July 1545 (EPICA: 9 July) and 30 April 1590 (EPICA: 28 April). For each match, the event page of the European AHEAD archive gave the studies with intensities by place, including those EPICA does not use (for example Alexandre and Alexandre, 2018, for 1471).

The rule for "intensities by place": a study with at least two named places. A single mention, in Kiev or Novgorod, is not enough. From 1900, "measured" means quality A, B or C in the Romanian catalogue, or any entry from 1980 on: the entries for 984–1979 come from Constantinescu and Mârza's 1980 compilation, the later ones from INFP's own determinations; entries from 1900–1979 graded D form the "weak parameters" class. The drawing goes down to magnitude 4; the headline count stays at magnitude 7.

The quake of 18 November 2026 is from USGS (us7000tyw8). No magnitude before 1900 has an hour or a place on the drawing; the uncertainty EPICA gives (±0.3 for those kept, up to ±0.5 for those classed otherwise) is in the notes. For 20 of the 37 magnitude-7 quakes the page names a written witness or a measurement; for the rest, the note says none was found.