ABSTRACT OBJECTIVE To describe urinary cotinine levels in tobacco farmers. METHODS A cross-sectional study was conducted in 2,570 tobacco farmers. All participants that reported green tobacco sickness in the week prior to the interview plus a subsample of 492 pesticide applicators were included. We collected urinary samples and information about sociodemographic, behavioral, dietary, occupational characteristics, and pesticide poisoning during their lifetime. Stratification by sex and smoking was performed and the Wilcoxon and Kruskal-Wallis non-parametrical tests were used to analyze cotinine means. RESULTS This study included 582 individuals. There was no difference in urinary cotinine means between green tobacco sickness symptomatic and asymptomatic individuals. Among non-smokers, having picked tobacco in the previous week was associated with higher cotinine means in both genders. Cotinine levels were higher on the first day of symptoms and reduced exponentially with each day in female non-smokers. Male non-smokers had higher levels on the second day and a more gradual reduction. The cotinine level rose up to 15 cigarettes/day of consumption. CONCLUSIONS The urinary cotinine measures exposure to nicotine up to its saturation point; while green tobacco sickness, affected by tolerance, indicates nicotine poisoning. Strategies to reduce nicotine exposure in tobacco production are needed. Mechanization could be an alternative, as long as it overcame the challenge of irregular terrain and did not affect the quality of the leaf. More studies are needed to evaluate the chronic effect of nicotine exposure.