The seasonal pattern in hip fracture rates and its relationship to weather variables was studied using hospital admission data for New South Wales. There was a consistent seasonal pattern for hip fracture, with a trough in the summer and a peak in the winter, for the six years for which data were available: 1981, 1983, 1986, 1988, 1989 and 1990. The seasonal trends for all years were statistically significant (P < 0.01) in men and in women, and in people who were 75 years and over. The relationship between the monthly admission rates for hip fracture in Sydney and such weather variables as mean daily minimum temperature, mean cloud cover, number of days with strong wind, number of days of fog, number of days of mist and number of days with 0.1 mm or more rainfall in a month were studied by Poisson regression. The mean daily minimum temperature for each month was the single weather variable independently and consistently associated with the monthly rates of hip fracture in both younger and older people. For the first time, a seasonal pattern for hip fracture and its close association with monthly temperature in Australia has been demonstrated.