The purposes of this study were to identify how pregnant women learned about alcohol consumption and tobacco use, and to identify the relationship between learning, health beliefs and behaviours. Determining how pregnant women learned was based upon Tough's and Knowles' view of learning and consisted of identifying knowledge levels, resources utilized, advice given, time in learning, and initiators of learning episodes. The ex post facto research design involved one-hour interviews with 128 primigravidas at 8 hospitals in British Columbia, 75% of the sample consumed alcoholic beverages before becoming pregnant and these women reduced their intake by an average of 82%; 39% smoked cigarettes before becoming pregnant and these women reduced their cigarette smoking by an average of 52%. Drinkers were advised not to consume alcoholic beverages during pregnancy, whereas smokers were told by friends and family members that it was okay to smoke during pregnancy. Engagement in other-initiated learning episodes was found to be correlated with reduced alcohol intake (p less than or equal to .05); whereas, health beliefs were not correlated with reduced alcohol intake. Neither self-initiated nor other-initiated learning was associated with reduced cigarette smoking; however, perceived risk was associated with reduced cigarette smoking. Knowledge about smoking was associated with health beliefs, suggesting that learning may be indirectly related to smoking behaviours. This study should be replicated with a larger sample to determine the directionality of the association between learning, beliefs and behaviours.