We used the Swedish Family-Cancer Database to analyse the effects of birth order and family size on the risk of common cancers among offspring born over the period 1958-96. Some 1.38 million offspring up to age 55 years with 50.6 million person-years were included. Poisson regression analysis included age at diagnosis, birth cohort, socio-economic status and region of residence as other explanatory variables. The only significant associations were an increasing risk for breast cancer by birth order and a decreasing risk for melanoma by birth order and, particularly, by family size. When details of the women's own reproductive history were included in analysis, birth orders 5-17 showed a relative risk of 1.41. The effects on breast cancer may be mediated through increasing birth weight by birth order. For melanoma, socio-economic factors may be involved, such as limited affordability of sun tourism in large families. Testis cancer showed no significant effect and prostate cancer was excluded from analysis because of the small number of cases.