The relationship between smoking status and incident Alzheimer's disease (AD) was investigated in a random stratified sample of a biracial community in Chicago. Analyses are based on 1,064 persons (of 1,134 evaluated) who had data on smoking status, disease incidence, and key covariates such as apolipoprotein allele status. During a mean of about 4 years of follow-up, 170 persons met criteria for incident AD. Current smoking was associated with increased risk of incident AD (OR = 3.4, 95% CI = 1.4-8.0) compared to persons who never smoked. There was no apparent increase in risk of AD for former smokers compared to persons who never smoked (OR = 0.9, 95% CI = 0.5-1.7). Apolipoprotein E allele status modified this association in that former smokers with a upsilon4 allele were less likely to develop AD (p = 0.04) than those who never smoked. Former smokers also appeared to have a reduced risk of developing AD as their pack-years of smoking increased (p = 0.02)such that the odds of developing AD increased by 50% for every 10 years of smoking cessation (OR = 1.3, CI = 0.9-1.7). The results suggest that older people who currently smoke are more likely to develop AD compared to those who never smoked; the relation between those who used to smoke but quit and the risk of AD is complex and requires further research.