The purpose of this study was to compare pure eccentric and concentric isokinetic training with respect to their possible specificity in the adaptation of strength and morphology of the knee extensor muscles. Ten moderately trained male physical education students were divided into groups undertaking eccentric (ETG) and concentric (CTG) training. They performed 10 weeks of maximal isokinetic (90 degrees x s(-1)) training of the left leg, 4x10 repetitions - three times a week, followed by a second 10-week period of similar training of the right-leg. Mean eccentric and concentric peak torques increased by 18% and 2% for ETG and by 10% and 14% for CTG, respectively. The highest increase in peak torque occurred in the eccentric 90 degrees x s(-1) test for ETG (35%) whereas in CTG strength gains ranged 8%-15% at velocities equal or lower than the training velocity. Significant increases in strength were observed in the untrained contra-lateral leg only at the velocity and mode used in ipsilateral training. Cross-sectional area of the quadriceps muscle increased 3%-4% with training in both groups, reaching statistical significance only in ETG. No major changes in muscle fibre composition or areas were detected in biopsies from the vastus lateralis muscle for either leg or training group. In conclusion, effects of eccentric training on muscle strength appeared to be more mode and speed specific than corresponding concentric training. Only minor adaptations in gross muscle morphology indicated that other factors, such as changes in neural activation patterns, were causing the specific training-induced gains in muscle strength.