Recent investigations have highlighted associations between maternal smoking in pregnancy and antisocial behaviour in offspring, and suggested the possibility of a causal effect. We used data from the 1970 British birth cohort study (BCS70) to examine these links in a large, population-based sample studied prospectively from birth to age 16. We found a strong dose-response relationship between the extent of pregnancy smoking and childhood-onset conduct problems, but no links with adolescent-onset antisocial behaviours. Effects on childhood-onset conduct problems were as marked for girls as for boys, and were robust to controls for a variety of social background factors and maternal characteristics. Controls for mothers' subsequent smoking history modified this picture, however, suggesting that the prime risks for early-onset conduct problems may be associated with persistent maternal smoking—or correlates of persistent smoking—rather than with pregnancy smoking per se.