Published observational estimates of the effect of thiazide diuretics on osteoporotic fracture risk vary from a 70% reduction to a 60% increase but there have been no randomized controlled trials. The aims of this study were to use the technique of meta-analysis to attempt to resolve this conflict and to explore whether duration and/or dose of therapy has an effect on osteoporotic fracture risk. The data sources utilized were Medline and Excerpta Medica databases supplemented by reviews and back references. A total of 18 observational studies that looked at the relationship between diuretics and fracture were located, of which 13, involving 29,600 subjects, had extractable data on thiazides and fracture occurrence. Current thiazide users were protected against hip fracture (OR 0.82, 95% CI 0.73-0.91). Thiazide use of long duration may be protective (OR 0.82, 95% CI 0.62-1.08) but not short duration (OR 1.23, 95% CI 0.99-1.54). The size of this effect, which compares favorably to other interventions, indicates that a randomized controlled trial to resolve the problem of potential confounders and safety profile would require a minimum of 7000 person-years of observation in those at highest risk of fracture (women aged 80 or over) which is unlikely to be pursued at the present time. The results of this meta-analysis indicate that current thiazide users have a 20% reduction in fracture risk and that long-term use may reduce fractures by a similar amount.(ABSTRACT TRUNCATED AT 250 WORDS)