Forty chronic smokers participated in a multisession treatment program designed both as a further assessment of the self-pacing procedure reported by Lando and Davison (1975) and as a partial replication of earlier aversive conditioning procedures. Subjects were randomly assigned to self-pacing, aversive conditioning, or a nonaversive control. A significant overall treatment effect emerged favoring aversive conditioning. Results did not support either the effectiveness of self-pacing or the cognitive dissonance explanation advanced for the Lando and Davison (1975) findings. Unfortunately, subjects in all conditions tended to increase smoking, eliminating specific between-group effects by the time of a 6-month follow-up.