How the GSAF Collects and Verifies Incident Data

The Global Shark Attack File, maintained by the Shark Research Institute in Princeton, New Jersey, is the world's most comprehensive database of shark-human interactions. It contains records dating back to 1580 — nearly 450 years of documented incidents. Understanding how those records are compiled and what quality controls exist is essential for interpreting the data correctly.

How incidents are reported

The GSAF does not have a network of field reporters. Incidents are submitted by the public, by researchers, by coastguard agencies, and increasingly by automated media monitoring. Each submission triggers a review process.

The verification process

Every incident record goes through several validation steps. Investigators — many of them volunteer experts with backgrounds in marine biology, forensic pathology, or ocean safety — review photographic evidence, medical reports, and eyewitness accounts. The goal is to determine:

• Whether a shark was definitively responsible (vs. propeller, boat, or other marine animal)
• Whether the incident was provoked or unprovoked
• The probable species, if identifiable from wound patterns or witness accounts
• Precise location and date

Classification challenges

The GSAF's classification scheme — provoked, unprovoked, boating, sea disaster, questionable — is more complex than most media coverage suggests. "Provoked" does not mean the victim was reckless; it includes cases where the shark was inadvertently touched or stepped on. "Sea disaster" covers incidents following shipwrecks or aircraft crashes, which historically accounted for the highest casualty events.

Historical records

Pre-20th-century records are inherently less reliable. The GSAF maintains them with explicit uncertainty flags, but they create systematic challenges for trend analysis. Researchers working with the full dataset must account for dramatically different detection and reporting rates across different eras and regions.

What the GSAF doesn't capture

Incident data from certain regions — particularly parts of Southeast Asia and sub-Saharan Africa — is significantly under-reported due to lack of formal reporting infrastructure. The true global incident rate is likely higher than published figures, with the gap concentrated in lower-income coastal nations.