This study determined the intratester and intertester variability and reliability of goniometric measurements taken by four physical therapists on upper and lower extremity motions of normal male subjects. The same subjects were measured once weekly for four weeks by testers with varied experience in goniometry. Data were analyzed by analyses of variance with repeated measures. Intratester variation for all measurements was less than intertester variation. Further, intertester variation was less for the three upper extremity motions than for those of the lower extremity. These findings indicate the necessity for using the same tester when effects of treatment are evaluated. When the same tester measures the same movement, increases in joint motion of at least three to four degrees determine improvement for either the upper or lower extremity. When more than one tester, however, measures the same movement, increases in joint motion should exceed five degrees for the upper extremity and six degrees for the lower extremity to determine improvement.