The purpose of the present study was to compare the prediction error of boys' adult stature depending on the Khamis-Roche method and regression equations proposed by the present author. The study is based on longitudinal data gathered in Kraków in between 1976-1988 and in 1994 (138 healthy boys aged 6-24 years). The predicted value from 7 to 16 years was compared with the body height measured during the last examination which was considered to provide adult stature. The equations proposed by the author were based on multiple regression analysis assessing the strength of the linear relationship between attained final height and other potentially, predictive measures of adult stature. The equations, which take into consideration current height, body weight at a given age, midparent's stature and, additionally, sexual maturity from 13 years, had the greatest predictive value. In the Khamis-Roche method, a tendency to underestimate the predicted adult height of Polish boys appeared. Median absolute error was from 2.2 to 1.4 cm at the age of 7 to 16 and was smaller for most age groups in the equations proposed by me. The proposed equations may be used in the prediction of adult stature of healthy boys, since they are based on the most current data, which is of importance in secular trends and interpopulation differentiation.