The purpose of this study was to evaluate the reliability and variability of repeated measurements of isometric (IM) lumbar extension (LB EXT) strength made at different joint angles. Fifty-six men (age, 29.4 .+-. 10.7 years) and 80 women (age, 24.3 .+-. 9.1 years) completed IM LB EXT strength tests on 3 separate days (D1, D2, and D3). On D1 and D2, subjects completed two tests (T1 and T2) separated by a 20- to 30-minute rest interval. For each test, IM LBV EXT strength was measured at 72, 60, 48, 36, 24, 12, and 0.degree. of lumbar extension. Mean IM strength values, within-day reliability coefficients, and test variability over the seven angles improved from D1 to D2 (D1: mean, 160.0 to 304.1 N .cntdot. m, r = 0.78 to 0.96, SEE = 37.6 to 46.9 N .cntdot. m; D2:mean, 176.3 to 329.1 N .cntdot. m, r = 0.94 to 0.98, SEE = 29.0 to 34.4 N .cntdot. m). Mean strength values leveled off by D3 (174.5 to 317.0 N .cntdot. m). The most reliable test results showed that the IM LB EXT strength curves were linear and descending from flexion to extension and ranged from 235.8 .+-. 85.2 to 464.9 .+-. 150.7 N .cntdot. m for men (extension to flexion) and from 134.6 .+-. 53.2 to 237.3 .+-. 71.9 N .cntdot. m for women. Lumbar extension strength was clearly greatest in full flexion, which is in contrast to previously reported results. One practice test was required to attain the most accurate and reliable results. These data indicate that repeated measures of IM LB EXT strength are highly reliable and can be used for the quantification of IM LB EXT strength through a range of motion.