Aims. We examined whether cigarette smokers in the United States can significantly reduce their smoking and maintain this reduction and, if so, whether this predicts an increase or decrease in the probability of smoking cessation in the future. Design. Longitudinal observation study. Setting. The 22 US cities of the Community Intervention Trial for smoking cessation (COMMIT). Participants. The 1410 subjects who smoked at both baseline and at 2‐year follow‐up. Intervention. Public health efforts to prompt cessation in half the communities. Measurements. Self‐reported cigarettes/day and abstinence at baseline, 2‐year and 4‐year follow‐ups. Findings. At 2‐year follow‐up. 60% of the subjects had either not changed or increased their smoking, 17% had decreased their smoking by 5‐25%, 15% by 24‐49% and 8% by 50%. Among the 40% who had reduced 5% at 2‐year follow‐up, 52% reported maintaining that reduction at 4‐year follow‐up. Reduction in smoking at year 2 did not prospectively predict an increase or decrease in the probability of making a quit attempt; nor did it predict eventually quitting by year 4. Conclusions. A substantial minority of US smokers are able to reduce their smoking and maintain this for long periods of time. Smoking reduction neither promotes nor undermines cessation.