Objective: We designed this study to assess the validity and reliability of pictogram for estimating body mass index (BMI). Study Design and Setting: Participants of Golestan cohort study during 2000-2004 were recruited in this study. Demographic and anthropometric information (weight, height, and BMI) were collected on all participants. A set of drawings (pictogram) ranging from very lean to obese were used to assess the individual's perception of their body size. Sensitivity and specificity of each pictogram score were calculated and cutoff points were determined using sensitivity/specificity plots. We used receiver operating characteristic curves to assess the validity of pictogram scores. Results: Of the 15,437 subjects enrolled in the study, 6,574 (42.6) were males and 8,863 (57.4) were females. Their mean±standard deviation age was 52.58±9.28 years. Pictogram scores 1, 2, and 3 were assigned to normal participants; pictogram score 4 was selected by overweight subjects, and finally, pictogram scores equal or higher than 5 were selected by obese ones (area under curve: 0.83-0.85). Conclusion: According to our results, pictogram is a valid measure for discriminating obese or overweight from normal individuals, and for distinguishing obese from overweight or normal individuals. So it can be concluded that body image pictogram is valid for discriminating normal and obese individuals. © 2010 Elsevier Inc. All rights reserved.