We assessed the incidence of hip fracture and ecological correlates in residents of 14 communities in six countries of Southern Europe. Hip fracture cases were recorded prospectively in defined catchment areas over a 1-year interval. A retrospective questionnaire was used to assess ecological differences between communities. During a 1-year period of observation a total of 3629 men and women over the age of 50 years were identified with hip fracture from a catchment of 3 million. In all communities the fracture rate increased exponentially with age. There were large and significant differences between centres in the doubling time for hip fracture risk with age and in crude and age-standardized rates. Greater than 4-fold and 13-fold differences in age-standardized risk were found amongst men and women respectively. The lowest rates were observed from Turkey and the highest from Seville, Crete and Porto. Fractures were significantly more frequent among women than men with the exception of three rural Turkish centres. Indeed, in rural Turkey the normal female/male ratio was reserved. Variations in incidence between regions were greater than the differences within centres between sexes, and there was a close and significant correlation between incidence rates for men and those for women in the regions studied. Excess female morbidity increased progressively from the age of 50 years but attained a plateau after the age of 80 years, suggesting a finite duration of the effect of the menopause. The retrospective questionnaire completed by 80% of cases suggested that differences in incidence between the communities studied could not be explained by differences in gonadal status in women. In both men and women cross-cultural associations were found with factors related to age or socioeconomic prosperity, the majority of which disappeared after adjustment for age. We conclude that there are marked and sizeable differences in the incidence rates of hip fracture throughout Southern Europe. The reasons for these differences are not known but affect both men and women, and are likely to be related to lifestyle or genetic factors rather than to differences in endocrine status.