Self-report is often used in large-scale studies. Therefore, it is important to determine the accuracy of self-report. In the Blue Mountains Eye Study, a population-based study of 3,654 older, community-dwelling Australians, subjects were asked about fracture history at the 5-year follow-up interview. All reported non-rib and vertebral fractures were radiologically confirmed. Hospital radiology records were searched for fracture records of people who reported no fractures in the 5-year period. Of 2,326 subjects who came to the interview, 272 subjects reported 318 fractures sustained since 1990. Overall, 34.6% of fracture reports could not be confirmed, mainly due to no record of treatment. Among self-reported fractures, false positive rates were 10.7% for all fractures and 4.8% for hip, 2.2% for wrist, 19.4% for ankle, and 6.6% for shoulder fractures. Sensitivity and specificity of fracture reports was high, with the lowest sensitivity for shoulder fractures (82.4%). Self-report of major osteoporotic fractures is reasonably accurate but may be improved by obtaining more details about treatment.