The purpose of this study was to investigate for possible relationships between knee flexion to extension peak torque ratios (F1/Ext(rat)) and low-back injuries in highly active males and females. Forty-eight male (age 25.9 +/- 4.5 years) and 41 female (age 27.3 +/- 2.6 years) competitive rowers, and 20 male (age 26.6 +/- 6.0 years) professional ballet dancers volunteered for the study. Each subject performed a test of lumbar and knee flexor flexibility, isokinetic dynamometry and completed a self-administered questionnaire. Flexibility was assessed by using the sit-and-reach test. Knee flexion to extension peak torques were bilaterally monitored at the angular velocities of 1.04 and 4.19 rad x s(-1). The questionnaire was designed to obtain information regarding the number of days off action (e.g., training, competition, and rehearsals), due to low-back injuries, for the 12-month period prior to testing. Results revealed significant negative correlation coefficients between knee F1/Ext(rat), obtained at 1.04 rad x s(-1), and days off physical activity for oarsmen (r = - 0.69; p < 0.01), oarswomen (r = -0.62; p < 0.01) and male dancers (r = -0.57; p < 0.05). No such correlations were found for either knee F1/ Ext(rat) obtained at the angular velocity of 4.19 rad x s(-1) or between the sit-and-reach test results and low-back injuries. A sub-group of 22 female rowers was re-tested after a 6-8 month period, during which a special hamstring strength training programme was introduced. The main conclusions were: a) the lower the F1/Ext(rat) the greater the degree of low-back injury, b) at least in female rowers, 6-8 months of hamstring strength training can contribute to a reduction of the incidence of low-back injury, and c) isokinetic assessment of quadriceps and hamstrings obtained at lower compared to higher angular velocities is more prognostic of low back injury.