To analyze archival physiotherapy records at a major military base in North Queensland, Australia, to investigate the epidemiology of injuries associated with sports and training, examining for possible risk factors for military training injury. A retrospective study was undertaken during a 62-month period, from 1987 to 1992, at Lavarack Barracks, Townsville, Australia, which services a dynamic population base of some 5,000 uniformed staff. Sociodemographic basic data, as well as treatment-related data (treatment area, number and type, interval between onset and initial treatment, reported cause), were used. Admission records were recoded according to the Orchard Sports Injury Classification System (version 2.0) standard. During the 62-month period from 1987 to 1992, 4,993 personnel, 96.2% (4,803/4,993) males and 3.7% (190/4,993) females, were referred for 5,025 physiotherapy treatments. The incidence of injuries requiring physiotherapy was 80.4 new patients per 5,000 personnel per month, and the incidence rate of injury was 19.3% per year or 0.19 injuries per person per year. The mean age of patients was 25.7 +/- 6.2 (SD) years, and the median age was 24 years, with a range of 17 to 59 years. Injuries were related to military training (29.3%, 1,471/5,025), diverse causes (21.2%, 1,072/5,025), sports (13.8%, 694/5,025), insidious onset (11.8%, 589/5,025), football (11.7%, 586/5,025), manual handling (4.2%, 211/5,025), motor vehicle crashes (4.1%, 206/5,025), and surgery (3.9%, 196/5,025). The four major body areas treated by physiotherapists were the knee joint (37.0%, 1,321/3,612), lumbar spine (29.8%, 1,075/3,612), ankle (19.9%, 719/3,612), and shoulder joints (13.8%, 497/3,612), which accounted for nearly three-quarters of all admissions. Of these, most were referred without definitive diagnosis (71.1%, 2,572/3,612), with the remainder comprising joint injuries (17.5%, 634/3,612), other types of pathology such as chest infections or neurological involvement, soft-tissue injuries (3.5%, 128/3,612), and bone damage (1.0%, 38/3,612). Most injuries (59.0%, 2,959/5,019) occurred during the 6 months between April and September referred to as the winter season, during which 71.8% of all football and 66.8% of all sports-related referrals were made. Significant associations were found between gender and injury cause (p < 0.001), gender and injury type (p < 0.01), body area affected and injury type (p < 0.00001), body area affected and injury cause (p < 0.00001), injury cause and injury type (p < 0.00001), and season and injury cause (p < 0.00001). Pretreatment interval was significantly associated with cause of injury (p < 0.00001), body area affected (p < 0.0001), and type of injury (p < 0.0001). Total number of consecutive treatments provided was significantly associated with both body area affected (p < 0.05) and injury type (p < 0.001). This study has used archival physiotherapy records for the purpose of exploring injury reporting patterns associated with a military population. The incidence profile for injuries using physiotherapy admissions is likely to be conservative because the patients are a group of injured military personnel selected by medical officers for physiotherapy treatment. This selection process needs further study, particularly because the majority of injuries referred for physiotherapy treatment are undiagnosed. This may be attributable in part to the cumulative and diverse nature of some injuries. Injury prevention needs to focus on activities relating to military training and football and other sports. Improved systems for recording detailed and accurate physiotherapy admission, treatment, and follow-up information are needed.