The objective was to describe the relationship between preoperative delay, postoperative complications, and risk of death at 6 months. The population is constituted of 200 subjects aged 65 years or older who were living at home and treated for a hip fracture in any of three of Québec's hospitals between April, 1st, 1987 and March, 31, 1989. Chi-square or F-test, and linear and logistic regression were used to test the relationship between the variables. Preoperative delays varied from 2 to 403 h (median, 45 h). Variations between hospitals were particularly important; median delay at hospital 1 was 109 h, at hospital 2, 36 h, at hospital 3, 30.5 h. Only 5% of the variance of the delay was explained by the subjects' characteristics before the fracture. The relationships between delay and postoperative complications are not significant. However, the risk of death at 6 months increased with the length of operative delay; the observed increase tends to be linear (p = 0.03). These results suggest first, that surgery for hip fracture had to be consider as an urgency, second, that it could be done 36 hours or less after the arrival at hospital.