BACKGROUND: Few studies have assessed the test-retest reliability of the isokinetic strength at 90 • /s and the isometric parameters for knee extensors. OBJECTIVES: To assess the reliability of isokinetic and isometric parameters of knee extensors, and to determine the smallest real difference between the test-retest measures. METHODS: Knee extensor strength was measured twice, 4 to 5 days apart, using a Biodex dynamometer in forty subjects (12 men, 28 women). The protocols consisted of 5 concentric contractions at 90 • /s and 5 isometric contractions. RESULTS: For women, test-retest reliability was very high for isometric and isokinetic peak-torque (ICCc,1 > 0.9 ), moderate for work (ICCc,1 = 0.82) and questionable for angle of peak torque (Angle-PT), mean Angle-PT and rate of torque development (RTD0−100 ms) (ICC < 0.67). The measurement errors were small for all variables, the SEM% ranged between 3.1% to 7.4%, and SRD% from 8.6% to 19.9%, except for RTD0−100 ms that was the most variable (SEM% = 16.4%; SRD = 47.5%). No differences were found in Angle-PT and RTD0−100 ms with regard to sex. CONCLUSIONS: The isokinetic values obtained at 90 • /s and isometric peak-torque for knee extensors are highly repeatable with a standardized test protocol and the SRD values can be used to detect real changes. Alternative strategies of analysis should be developed to improve reliability of Angle-PT and RTD0−100 ms. Angle-PT and RTD0−100 ms do not appear affected by gender