We compared the effects of concentric (Con) and eccentric (Ecc) isokinetic training on quadriceps muscle strength, cross-sectional area, and neural activation. Women (age 20.0 +/- 0.5 yr) randomly assigned to Con training (CTG; n = 16), Ecc training (ETG; n = 19), and control (CG; n = 19) groups were tested before and after 10 wk of unilateral Con or Ecc knee-extension training. Average torque measured during Con and Ecc maximal voluntary knee extensions increased 18.4 and 12.8% for CTG, 6.8 and 36.2% for ETG, and 4.7 and -1.7% for CG, respectively. Increases by CTG and ETG were greater than for CG (P < 0.05). For CTG, the increase was greater when measured with Con than with Ecc testing. For ETG, the increase was greater when measured with Ecc than with Con testing. The increase by ETG with Ecc testing was greater than the increase by CTG with Con testing. Corresponding changes in the integrated voltage from an electromyogram measured during strength testing were 21.7 and 20.0% for CTG, 7.1 and 16.7% for ETG, and -8.0 and -9.1% for CG. Quadriceps cross-sectional area measured by magnetic resonance imaging (sum of 7 slices) increased more in ETG (6.6%) than in CTG (5.0%) (P < 0.05). We conclude that Ecc is more effective than Con isokinetic training for developing strength in Ecc isokinetic muscle actions and that Con is more effective than Ecc isokinetic training for developing strength in Con isokinetic muscle actions. Gains in strength consequent to Con and Ecc training are highly dependent on the muscle action used for training and testing. Muscle hypertrophy and neural adaptations contribute to strength increases consequent to both Con and Ecc training.