The prevalence of varicose veins has been studied in several population-based samples in the South Pacific. There is a striking gradient in the age-standardized prevalence rates (0 x 8 per cent to 43 x 7 per cent) with atoll dwellers having the lowest rates, Rarotongans intermediate rates, and New Zealand Maoris and Pakehas the highest rates. The relationship between varicose veins and age, sex, height, weight, body mass and parity is examined and it is concluded that none of these factors are responsible for the observed gradient. The consumption of refined carbohydrates follows the same gradient and the possible aetiological role of this factor is discussed.