We hypothesized that training with eccentric contractions only (therefore using higher loads) would yield greater muscle structural and strength gains compared with conventional resistance training. Nine older adults (mean +/- s.d. age, 74 +/- 3 years) were assigned to a conventional (CONV) resistance training group performing both concentric and eccentric contractions and 10 (age, 67 +/- 2 years) to an eccentric-only (ECC) resistance training group. Both groups trained three times per week for 14 weeks at 80% of the five-repetition maximum, specific to each training mode. Maximal knee extensor torque was assessed during isometric, concentric and eccentric contractions across a range of angular velocities (0-3.49 rad s(-1)). Vastus lateralis muscle architecture (fascicle length, pennation angle and muscle thickness) was assessed in vivo at rest using ultrasonography. Training increased fascicle length in both groups, but the increase was significantly greater in the ECC (20% increase) than the CONV group (8% increase). Conversely, pennation angle significantly increased in the CONV (35% increase) but not in the ECC group (5% increase). Muscle thickness increased to a similar extent in both groups (approximately 12% increase). In the ECC group, eccentric knee extensor torque increased by 9-17% across velocities, but concentric torque was unchanged. Conversely, in the CONV group, concentric torque increased by 22-37% across velocities, but eccentric torque was unchanged. Instead, isometric torque increased to a similar extent in both groups (approximately 8% increase). Thus, the two training regimens resulted in differential adaptations in muscle architecture and strength. These results suggest that the stimulus for adding sarcomeres in-series and in-parallel may be different, which implies that different myogenic responses were induced by the two different training methods.