The purpose of this investigation was to study the effect of one-legged exercise on the strength, power and endurance of the contralateral leg. The performance of the knee extensor and flexor muscle of 20 healthy young adults (10 men and 10 women) was first tested by Cybex II+ and 340 dynamometers. Then 10 subjects were chosen at random to train using one leg three times a week for 7 weeks whilst the other 10 served as controls. During the 8th week, the tests were repeated. Both quadriceps and hamstring muscles of the trained subjects showed a cross-transfer effect from the trained limb to the untrained side. This concerned the strength and power, as well as endurance characteristics of these muscles. The average change in peak torque of the quadriceps muscle was +19% (P less than 0.001) in the trained limb, +11% (P less than 0.01) in the untrained limb and 0% in the control limbs. In hamstring muscles the changes were +14% (P less than 0.01), +5% and -1%, respectively. Concerning muscle endurance (work performed during the last 5 contractions in the 25-repetition test) the corresponding changes were +15% (P less than 0.01), +7% (P less than 0.01), and -1% in quadriceps muscle, and +17% (P less than 0.05), +7%, and -3% in hamstring muscles. The average strength benefit in the untrained limb was +36% (hamstring muscles) and +58% (quadriceps muscle) of that achieved in the trained limb. Untrained hamstring muscle showed better benefits in the endurance parameters than in strength or power parameters, while in the quadriceps muscle this effect was reversed. A positive relationship was observed between the changes (greater improvement in the trained limb resulted in greater improvement in the untrained limb) (hamstring muscles: r = 0.83, P less than 0.001, quadriceps muscle: r = 0.53, P less than 0.001). In endurance parameters, this relationship was almost linear while in the strength and power parameters the results were more in favour of a curvilinear relationship with limited benefit.