Data were analysed using Statistical Package for the Social Sciences (SPSS) 16.0 and Microsoft Office Excel 2003 for Windows. Data on the eight variables on victims, injuries and circumstances were organised in cross-tabulations and association between variables was tested with Pearson’s Chi square test. Data on breeds were presented in contingency tables with the variables breed (breedx or nonbreedx) and population (biting population or canine population). To test for statistical significance of differences between the two fractions, we used Fisher’s Exact test for breeds and Pearson’s Chi square test (without Yates’ correction) for breed groups. Responses in the category ‘unknown’ were considered as missing data in statistical analyses. Tests were two tailed, and significance was accepted if P < 0.05.