The use of self reported height and weight is very common in epidemiologic studies and clinical settings. In this study we assessed the accuracy of selfreported weight, height and body mass index by comparing with measured data. Data were obtained from 1065 persons(657 men, 358 women) who visited a health care center for routine medical check-up during the period September 1994-February 1995. The correlation between reported and measured height and weight was very strong. There were some discrepancies between two data, however, and men and women differ ed somewhat in their pattern of misreporting. Weight was overestimated by 0.27kg in men and underestimated by 0.09kg in wo men whereas height was overestimated in men(0.88cm) and women(2.15cm). These misreported data resulted in an underestimation of body mass index and low sensitivity for diagnosing overweight in men and women. The size of error was larger in women than in men. So it suggests that the self reported height and weight data could be used as a continuous variable with little error, but misclassification could be occured when the self reported data were used as a categorical variable. Errors in self reported weight and height, BMI were related to a person's over-weight status. The more people overweighed the more they underestimated their weight. Educational level and age, exercise, marri age had no relationship with the accuracy of self reported data. In conclusion, the use of self reported weight and height in epidemiologic studies and clinical settings could result underestimation of obesity, especially in obese female group.