OBJECTIVES: Osteoporosis therapy places extensive behavioral demands on patients, including the need for precisely scheduled daily or weekly medication intake, regular physical exercise and frequent visits to health care providers. Although effective treatments are available, compliance with osteoporosis therapy is known to be low. It has previously been demonstrated that medication non-compliance in actual practice -defined as taking less than 80% of prescribed medications - significantly increases the risk of fractures. The objective of this study was to explore the effect of varying the definition of non-compliance to identify the minimum level of compliance required to reduce fracture risk. METHODS: Demographic, prescription drug use, physician services and hospitalization information for women with osteoporosis who were dispensed an osteoporosis medication between 1996 and 2001 was obtained from the Saskatchewan Health data files. Fractures were identified by hospitalization or physician visits with a relevant diagnostic or procedure code. Different levels of compliance were defined based on the proportion of time during which patients had drugs available: <50% (poor), 50-80% (medium), 80-90% (good), and >90% (excellent). A Cox proportional hazards model was used to assess the effect of compliance, defined as a time-dependent covariate, controlling for other known risk factors for fractures. Patients were studied for up to three years following the index prescription. RESULTS: Among 11,249 women suffering from osteoporosis the overall fracture rate was 4.8 fractures per 100 person-years. Poor compliance led to a 40% increase in fracture risk (95% CI: 19% - 66%) compared to excellent compliance, while medium compliance increased risk by 20% (95% CI: 0% - 43%). No statistically significant differences in fracture risk were observed, however, between good and excellent compliance. CONCLUSION: A high level of compliance (80% - 100%) is required to reduce osteoporosis-related fracture risk.