In a 24-year sub-sample taken from a 42-year period of study (1950-1991), hip fracture incidence was analysed from a defined catchment area within one hospital. During this time, 8,256 hip fractures occurred in a generated risk population of 1,915,571 person-years. Crude incidence increased three-fold in women and five-fold in men. In men, the age-specific increase was twice as large as the age drift. In women, the two components were of equal size. The more marked increase in men caused the female:male ratio to decrease from 4.2 in 1950 to 2.4 in 1991. In men, all age classes experienced a significant yearly increase (1.6% in the 50-59 age group, 3.9% over the age of 80). In women, only the 70-79 and 80+ age groups showed a significant increase (1.4%, 2.3%). In the age-standardised curve, a levelling off occurred during the mid-80s. In women, this was attributable to changes in climate during wintertime. In men, no significant association was found with temperature. The age-standardised curve followed an approximate linear trend with an increase of 6.4/100,000/year in women and 4.9/100,000/year in men. The cumulative rate for the age group 50-79 years doubled in men but increased only by one-third in women. The impact of increasing incidence in men compared with women is discussed using an osteoporosis model consisting of base risk, senile risk, and post-menopausal risk.