To assess whether long-term thiazide use is associated with a decreased risk of hip fracture, a nested case-control study was done in the Canadian province of Saskatchewan between 1984 and 1985 among residents who were 65 years of age or older and who were not receiving other drugs thought to affect bone mass. There were 905 hip fractures identified from hospital discharge records and 5137 population controls matched for age, sex, and calendar year. Drug use was ascertained from computerised pharmacy records. Risk of hip fracture decreased significantly with increasing duration of current thiazide use: relative risk (95% confidence interval) of 1.2 (0.9-1.5) for less than 2 years use, 0.8 (0.7-1.0) for use of 2-5 years, and 0.5 (0.3-0.7) for 6 or more years. In contrast, there was no such trend for use of other antihypertensive-diuretic drugs (relative risk 0.9 [0.6-1.3] for use of 6 or more years). This protective effect was not altered by age, sex, nursing home residence, previous hospital admission, or use of other antihypertensive-diuretic drugs or psychotropic drugs. Medical record review for a sample of 235 cases suggested this finding was not due to confounding by body mass, ambulatory status, functional status, or dementia. These results support the hypothesis that thiazides protect against osteoporosis in elderly people.