The outcome of snakebite is related to the biting species but it is often difficult to identify the biting snake, particularly in community settings. We have developed a clinical scoring system suitable for use in epidemiological surveys, with the main aim of identifying the presumed biting species in those with systemic envenoming who require treatment. The score took into account ten features relating to bites of the five medically important snakes in Sri Lanka, and an algorithm was developed applying different weightings for each feature for different species. A systematically developed artificial data set was used to fine tune the score and to develop criteria for definitive identification. The score was prospectively validated using 134 species-confirmed snakebites. It correctly differentiated the bites caused by the three snakes that commonly cause major clinical problems (Russell's viper (RV), kraits and cobra) from other snakes (hump-nosed viper (HNV) and saw-scaled viper (SSV)) with 80% sensitivity and 100% specificity. For individual species, sensitivity and specificity were, respectively: cobra 76%, 99%; kraits 85%, 99%; and RV 70%, 99%. As anticipated, the score was insensitive in the identification of bites due to HNV and SSV. (C) 2005 Royal Society of Tropical Medicine and Hygiene. Published by Elsevier Ltd. All rights reserved.