Piracy never goes away. It moves. I mapped 7,511 attacks on ships from 1993 to 2020 and followed the hotspots as they shifted: from the South China Sea and Indonesia to the Strait of Malacca, then to the Gulf of Aden and the open Indian Ocean, then to the Gulf of Guinea, and, in 2025, back to the Singapore Strait.
When people picture pirates today they picture Somali skiffs, ladders against the side of a tanker, a ship held for ransom off the Horn of Africa. That was real, and for four years it dominated the world's piracy. But it was one hotspot among several, and it came and went. The Crime at Sea database, built by Benden, Feng, Howell and Dalla Riva from the reports of the International Maritime Bureau and the US National Geospatial-Intelligence Agency, gives a position for 7,511 attacks and attempted attacks from 1993 to 2020. Put on a map and played year by year, it shows piracy drifting around the world's shipping lanes like weather.
The database stops in 2020. To bring the story up to today I add the International Maritime Bureau's own headline counts for 2021 to September 2026. Those are totals and regional counts from its press releases, not mapped points, and I keep them apart on this page.
Each dot is one reported attack, coloured by what happened: boarded, hijacked, fired upon, or attempted and fought off. The hexagons show where attacks crowd together in the years you choose. Press play to watch the hotspots move year by year, pick an era, or drag the slider. Click a dot for the record and, from 2015, the IMB's own account.
Dots: 7,511 attacks and attempted attacks, 1993 to 2020, from the Crime at Sea database (1993 to 2014 from the MAPPING piracy data, which compiles IMB and NGA ASAM reports; 2015 to 2020 from IMB reports). Positions are as reported; some anchorage attacks share one position. Hexagons count the attacks shown, at the current zoom. "Fired upon" is recorded almost only from 2015. Shipping lanes (under Map layers): World Bank and IMF Global Shipping Traffic Density, AIS 2015 to 2021, summed to 1 degree squares. Outlines: Natural Earth. Base map tiles © Esri.
Reported attacks rose from 95 in 1994 to a peak of 469 in 2000, climbed back above 400 a year from 2009 to 2011 (1,287 in three years), and then fell to 190 in 2020. The International Maritime Bureau's counts since then sit between 115 and 137 a year. But the total hides the real story, which is where the attacks happened.
Pirate attacks per year, by region, 1993 to 2026
Mapped attacks to 2020, coloured by region. From 2021: IMB headline totals (hatched), which are not split by region here. 2026 is January to September only. Hover a bar.
In the 1990s piracy was above all an East and Southeast Asian problem. From 1993 to 1999 the South China Sea, the Philippines and East Asian waters saw 332 attacks, 26% of the world's, more than any other region, with ships boarded off Hong Kong, in the Gulf of Thailand and around the Philippines. Indonesia was close behind and overtook it as the decade ended: attacks there peaked at 105 in 2003, many of them robberies of ships at anchor off Jakarta, Balikpapan, Samarinda and Belawan. Across 1993 to 2004 Southeast and East Asian waters, from the Malacca and Singapore straits to the South China Sea and Indonesia, held 52% of all reported attacks.
The Strait of Malacca, the narrow lane between Sumatra and Malaysia that carries much of the trade between Europe, the Gulf and East Asia, peaked at 104 attacks in 2000. From 2000 to 2004 it averaged 51 a year. Then it fell: 18 in 2005, and an average of 14 a year from 2006 to 2020, with no year above 32. The fall came with coordinated sea patrols by Malaysia, Singapore and Indonesia from 2004 and joint air patrols from 2005, and after the 2004 Indian Ocean tsunami devastated the Aceh coast.
Then the hotspot jumped west. Attacks in Somali waters, the Gulf of Aden, the southern Red Sea and the Arabian Sea went from 53 in 2007 to 231 in 2011. Over 2008 to 2011 they were 48% of all attacks in the world, and 163 of the 197 hijackings (83%). These were different attacks from the anchorage thefts of Asia: 99% were on ships under way, often far out to sea, and most failed: 73% were attempts that did not get on board. Then they collapsed, by 92% between 2011 and 2013, as naval patrols, armed guards, protective measures on ships and action against pirate groups ashore took effect. From 2014 to 2020 the database records only 3 hijackings in these waters, all in 2017.
West Africa was never quiet: the anchorages off Lagos are the fourth busiest square in the whole record. But as Somali attacks vanished, the Gulf of Guinea became the world's leading hotspot. It peaked at 83 attacks in 2018 and still had 82 in 2020, 43% of the world's total that year, up from 8% in the 1990s. The attacks moved further out: the median distance from shore rose from 7.2 km before 2016 to 23.2 km in 2016 to 2020, as gangs from the Niger Delta took crews off ships far offshore and held them for ransom: 60 of the 331 reports from 2016 to 2020 mention kidnapping. 49 of the 69 ships fired upon from 2016 to 2020 were there.
The Singapore Strait, the busy channel between Singapore and Indonesia's Batam and Bintan islands, had few attacks before 2008 (about 6 a year). It surged to 81 in 2015; 89% of the attacks in 2014 and 2015 were boardings rather than attempts or hijackings. It then dropped to 8 in 2016. In 2025 the IMB counted 80 incidents there, about as many as the 2015 peak and 58% of the world's reported incidents (see section 06).
The busiest pirate hotspots of each era
Attacks in each 1 degree square (about 110 km across), the top five per era, coloured by region. Click a row to see it on the map.
The single busiest square across all 28 years is the Chittagong anchorages, Bangladesh, with 366 attacks, most of them thefts from ships waiting at anchor. The twenty busiest squares, out of 1,093 with any attack, hold 35% of all attacks. The Somali attacks are the exception: they were spread across an ocean, so even at their peak only two Gulf of Aden squares reach the top five.
Most piracy is not dramatic. 64% of all the attacks were boardings, usually by a few robbers who climb aboard, take whatever they can carry, and leave. 27% were attempts fought off or abandoned. Hijackings, where the attackers take control of the ship, were 7% (511), peaking at 53 in 2010. Of the attacks where the ship's state is known, 60% were on ships at anchor or alongside a berth.
Attack types by era
Share of attacks in each era. "Fired upon" is recorded almost only from 2015, when the database switches to IMB reports; before that, ships fired upon but not boarded are mostly counted as attempted. 1993's 101 attacks have no type. Hover a bar.
Half of all attacks happened within 10.1 km of land, and 65% within 20 km. Ships at anchor were attacked a median 5.2 km from shore. The Somali attacks broke that pattern. With motherships towing skiffs, Somali pirates hunted ships hundreds of kilometres out: the median distance of attacks in these waters was 131 km in 2008 to 2011, reaching 204 km in 2010, and 41% were more than 100 nautical miles (185 km) from land. Of all the attacks that far out, 72% were in Somali waters.
Distance from the nearest shore, every attack, 1993 to 2020
Each dot is one attack (log scale). Red: Somali coast, Gulf of Aden, Red Sea and Arabian Sea. Lines: median per year for those waters (red, years with five or more attacks) and for everywhere else (dark). Hover a dot.
Pirates go where the ships are, and where ships slow down or wait. Using the World Bank and IMF map of ship traffic from satellite and coastal AIS signals, summed to 1 degree squares, 79% of attacks were in the busiest quarter of squares with any traffic, and 43% in the busiest tenth. But traffic alone does not make a hotspot. The very busiest 1% of squares, in the North Sea, the English Channel and the Baltic approaches, saw only 0.5% of attacks. Turn on "Shipping lanes" under Map layers to compare.
Attacks by how busy the sea is
1 degree squares with any ship traffic, ranked by AIS traffic density (2015 to 2021), against the share of all attacks (1993 to 2020) that fell in them.
The country indicators that come with the database tell a familiar story, with all the usual caution. 82% of attacks inside a country's waters were in countries whose income per person was below the world median that year, and 75% in countries scoring under 35 out of 100 on the Corruption Perceptions Index. That is a correlation, not a cause: poorer countries also have many of the world's busiest anchorages and straits.
The mapped data end in 2020. For 2021 to 2026 I use the International Maritime Bureau's published headline counts. They are counts of reported incidents, not positions, and they are not directly comparable with the mapped series, though the database itself was built largely from IMB reports.
Worldwide, the IMB counted 132 incidents in 2021 and 115 in 2022, the lowest for three decades, then 120, 116 and 137. In January to September 2026 it counted 51, 56% fewer than the 116 of the same months in 2025, mostly because the Singapore Strait went quiet. Indonesia had 12 incidents in 2025 and 6 in the first nine months of 2026, and the Philippines 9 boardings. So in 2026 the picture is two hotspots of very different kinds: petty theft in the busiest strait in the world, and the return of ransom hijacking off Somalia.
Headline counts, 2021 to September 2026
IMB reported incidents: world total and the regions the press releases name. Blank where a release does not give a number. Hover a bar.
Piracy never goes away. It moves to wherever ships slow down, the law is thin and somebody will pay.
The attacks. The Crime at Sea database (Benden, Feng, Howell and Dalla Riva, 2021) gives 7,511 attacks from 1993 to 2020 with a date, position, attack type, the ship's state (anchored, berthed or steaming), the nearest country and its exclusive economic zone, and the distance to shore. Attacks from 1993 to 2014 (6,338) come from the MAPPING piracy data compiled by Brandon Prins from IMB and NGA ASAM reports; those from 2015 to 2020 (1,173) come straight from IMB reports and include their text. I merged "Boarding" and "Boarded", which the source uses in alternate years for the same thing, and kept the positions as given.
Regions are simple geography: the Singapore Strait is a box from 103.4°E to 104.65°E and 0.85°N to 1.6°N (including the Batam and Bintan anchorages); the Strait of Malacca a polygon from the Andaman Sea to Singapore; then Indonesian waters, the South China Sea and East Asia, the Indian subcontinent (within 300 km of its coast), the East African coast (within 60 km of Kenya to South Africa), Somali waters (everything else between 32°E and 80°E, 15°S and 30°N, outside the Persian Gulf), West Africa (30°W to 16°E, 20°S to 20°N) and the Americas. Hotspots are 1 degree squares. Shipping comes from the subsampled overview of the World Bank and IMF density raster, so I only use it summed to 1 degree.
What this is not. These are reported attacks. Many are never reported, especially thefts from small and local craft and attacks on fishing boats; reporting also changes with insurance, politics and who is watching. The 1993 to 2014 records and the 2015 to 2020 records were coded by different people, so small shifts at 2015, such as "fired upon", are partly coding. The IMB counts for 2021 to 2026 are headline numbers from press releases, not mapped points.