The study aim was to quantify physiotherapy service distribution among compensated workers with musculoskeletal disorders, and identify risk factors for under- and overuse. Andersen and Newman s model of service use determinants was adapted for a compensated population, to provide a conceptual basis for the analyses. Methods WorkSafe Victoria (Australia) workers compensation claims were analysed retrospectively. Workers with musculoskeletal disorders resulting in at least 10 days off work were included if their claim commenced between 1-1-2001 and 1-1-2005 (n = 36,995). Physiotherapy use over 4 years of follow-up was determined from service payment data. Regression models were used relating individual level predictors, regional physiotherapist supply and the role of individual physiotherapists to service use. Results Physiotherapy was used by 26,026 (70 ) workers. Young age, male gender, working as a labourer, disorders of the joints, and not being hospitalised were associated with non-use. Use above the 90th percentile (>125 sessions over 4 years) was considered high use : high users accounted for 41 of all use. Age 50-60, female gender, working as tradespersons, and substantial hospital costs were associated with high use. For workers living in the most disadvantaged areas, use was positively associated with supply. Negative binomial modelling of the role of physiotherapists indicated that service providers were associated with the number of sessions used. Conclusions Physiotherapy services were not underused, but a small group of patients had very high use. Recommendations to limit overuse should be aimed at physiotherapists, and these could include effective monitoring of adherence to proposed treatment plans.