Estimates of snakebite mortality are mostly based on hospital data, although these may considerably underestimate the problem. In order to determine the accuracy of hospital-based statistics, data on snakebite mortality in all hospitals in the Monaragala District of Sri Lanka were compared to data on snakebite as the certified cause of death for the district, for the 5-year period between 1999 and 2003. Data were cross-checked in a sample of hospitals and divisional secretariats within the district. Hospital statistics did not report 45 (62.5%) of the true number of snakebite deaths in the Monaragala District. Twenty-six (36.1%) of the victims either did not seek, or had no access to, a hospital. Another 19 (26.4%) had arrived at hospital, but had done so too late to receive treatment. Our study confirms the limitations of official hospital-based mortality data on snakebite. (C) 2005 Royal Society of Tropical Medicine and Hygiene. Published by Elsevier Ltd. All rights reserved.