OBJECTIVE To examine variations in rates of hysterectomy for the five main indications for the procedure in regions of Ontario. METHODS Cross-sectional population-based analysis of hospital discharge abstracts. METHODS All acute care facilities in Ontario. METHODS All 65,599 women whose hospital record contained a procedure code indicating that a hysterectomy was performed between Apr. 1, 1988, and Mar. 31, 1991. Duplicate cases, records of cancelled procedures and nonresidents were excluded. METHODS Crude and age-adjusted rates of hysterectomy, by indication, for each region of Ontario. RESULTS Five indications accounted for more than 80% of hysterectomies performed. The median age-adjusted rate of hysterectomy for Ontario regions during the study period was 6.25 per 1000 women, with a 2.7-fold variation among regions. The regions with rates of hysterectomy in the highest quartile tended to be rural, and those with rates in the lowest quartile tended to be urban areas with teaching hospitals. When rates of hysterectomy for specific indications were examined, they showed substantial variations among regions in the rate of the procedure for menstrual hemorrhage (18-fold variation), uterine prolapse (9.3-fold) and endometriosis (6.3-fold). A smaller but still significant variation was shown in the rate of hysterectomy for leiomyoma (2.3-fold). Regional variation in the rate of hysterectomy for cancer (2.5-fold) was not statistically significant. CONCLUSIONS There are large interregional variations in rates of hysterectomy, especially for indications that are more discretionary than others (i.e., menstrual hemorrhage, uterine prolapse and endometriosis) and less variation in rates when treatment options and diagnosis are clear-cut. This result suggests the need for more definitive practice guidelines on treatment of the indications for which the rate is more variable.