To determine the effect of pregnancy on prognosis in melanoma we investigated 388 women treated for stage 1 primary cutaneous disease during their childbearing years. 85 women had been treated before any pregnancy, 92 during pregnancy, 143 after they had completed all pregnancies, and 68 between pregnancies. Women who had received treatment while pregnant had primary tumours of significantly greater thickness than did those in the other three groups (p=0·002). Other possible confounding factors (site, age, parity) did not differ between the groups. Once tumour thickness was controlled for, survival rate of women in whom melanoma was diagnosed and treated while they were pregnant did not differ from that in the other three groups. Cox regression analysis showed no differences between the three groups of women who were not pregnant at diagnosis. Women with melanoma should be advised about pregnancy on the basis of thickness and site of tumour and evidence of vascular spread, and not hormonal status. © 1991.