The aim of this study was to describe the epidemiology of moderate and severe injury morbidity in a defined population on the basis of ambulance records, and to validate ambulance records as a potential source of surveillance. A geographical target area was defined; the county of Värmland, Sweden. All ambulance attendances and hospitalizations for unintentional and intentional injury in 2002 were selected, analysed, and compared. Ambulance data comprised 3,964 injury cases (14.5/1,000). Most injuries for which ambulance attention was sought occurred in road traffic areas (27%), followed by residential areas (20%), school and institutional areas (14%), and sports areas (8%). An ecological comparison between ambulance-based data and hospitalizations showed that ambulance services captured approximately the same amount of injury cases (3,235 ambulance reports, as compared to 3,456 hospital discharges) with a similar profile. This study provides epidemiological support for ambulance services as a potential source of regular surveillance data on moderate and severe injuries. However, at a population level, our results indicate that ambulance data tend to overestimate some injury categories, and underestimate others, as compared to hospital data. The significance of these differences for preventive work, as well as other practical aspects of the feasibility of regular injury surveillance, will be analysed and discussed on the basis of general criteria for evaluation of surveillance systems in a forthcoming paper.