Altogether, 100 uninjured professional rugby league players were evaluated over a 2-year period. Their height, body mass, sum of skinfolds, girths and bone diameters were recorded. A Cybex 340 isokinetic dynamometer was used to determine peak torque, work, power, endurance ratios and peak torque ratios of the hip abductors and adductors (5 repetitions at 0.52 and 2.08 rad. s(- 1); 20 repetitions at 3.66 rad. s(- 1)) and knee flexors and extensors (4 repetitions at 1.04 and 3.14 rad. s(- 1); 30 repetitions at 5.22 rad. s(- 1)). Hip abduction and adduction were also assessed with the hip in external rotation. Discriminant function analysis was conducted on all predictor variables to develop a multivariate predictive model capable of classifying players with a high degree of accuracy into groups with and without a groin injury. The model consisted of eight variables and correctly classified 91.7% of the non-injured players and 90.5% of the injured players. The correct classification for the model as a whole was 91.4%. The aetiological factors identified as being related to injury of the groin musculotendinous unit included abduction and adduction-with-rotation peak torque, angle of adduction and abduction-with-rotation peak torque, strength ratio of hip muscle groups, bilateral difference in extension peak torque, femur diameter and body mass.