Forty normal subjects, 19 males and 21 females, between the ages of 16 and 43 years were studied to examine the reliability of Cybex knee extension curves utilizing Segmental Curve Analysis (SCA). Each subject performed a standardized isokinetic knee extension/flexion test on the Cybex II isokinetic dynamometer. Test protocol consisted of 5 maximum repetitions at a speed of 60 deg/sec. Within one week of the initial test, each subject performed a retest. During testing, all torque and angle measurement information from the Cybex was transmitted to the SCA system. The SCA system plotted, analyzed, and quantified each torque curve for seven specific parameters. The following parameters were quantified by the SCA system: (1) torque at 20 degrees, 70 degrees, and peak torque of knee extension; (2) area of the curve to 20 degrees, 70 degrees and peak torque of knee extension; and (3) area of the torque curve between 20 and 70 degree angles of knee extension. The parameters of peak torque (r=.83 to .97; R=.98 to .99), area to 70 degree angle of knee extension (r=.76 to .87; R=.96 to .98), and area between 20 and 70 degree angles of knee extension (r=.75 to .92; R=.97 to .99) appeared to be the most reliable across trials and days, and were also found to elicit the least amount of variation for both male and female subjects. Coefficients of variation on the parameters of peak torque, area to 70, area between 20 and 70 for females ranged from 15% to 18%. For males, the coefficients of variation for peak torque, area to 70, and area between 20 and 70 ranged from 23% to 27%. Variables quantifying torque and power indices in the middle segment of torque curves appeared to be most reliable as analyzed by the SCA system.