Outcome expectancies and self-efficacy are the two main predictors of smoking behavior in social learning theory. There has been considerable research effort on self-efficacy, but almost nothing on outcome expectancies related to smoking. Rohsenow, Abrams, Colby, Gulliver, Niaura, and Monti constructed the Smoking Effects Questionnaire to measure such expectancies. In a questionnaire survey among a random sample from the adult Norwegian population, 1,639 out of 5,014 respondents were current smokers. Confirmatory factor analyses were applied to examine the questionnaire's dimensionality. The fit of the original factor structure was compared to the structure which emerged from a principal component analysis and was well suited to the Norwegian data. Also, attitudes about how smoking affects other people and a general attitude towards smoking were measured. The outcome expectancies together with the specific attitudes predicted the general attitude, while the opposite was not true. LISREL-analyses indicated that women were more influenced by negative psychological effects and reduction of negative affect in their general attitude, while men were more influenced by negative physical effects.