National assault injury surveillance has identified major seasonal variation, but it is not clear whether assault injury is a seasonal problem in large cities. Relationships between community violence, calendar events and ambient conditions were investigated with reference to prospective, Accident and Emergency (A&E) derived information obtained from people injured in assaults in Cardiff between 1 May 1995 and 30 April 2000. Records of daily local ambient conditions included data relating to temperature, rainfall and sunshine hours and data of major local sporting events and annual holidays were studied. Pearson correlation coefficients were used to evaluate associations between variables. Overall, 19,264 assault-related A&E attendances were identified over the 5-year period. Almost three-quarters were males. Violence was clustered predominantly on Saturdays and Sundays, New Year and rugby international days. Temperature, rainfall and sunlight hours did not correlate significantly with violence (P > 0.05). The findings indicate that injury reduction effort should be intensified at the known risk times for violence and that in a capital city/regional centre violence cannot be predicted on the basis of ambient conditions.