This study investigated effects of downhill (DR) versus level (LR) running training on various muscular and aerobic performances. Eighteen healthy young males conducted either DR (DR group (DRG), n = 10: -10% slope) or LR (LR group (LRG), n = 8) training at a target heart rate (HR) associated with lactate threshold (LT) for 20 min·session, 3 sessions·week, for 5 weeks. Before and after the interventions, the following variables were measured: knee extension torque (-150, -30, 0, 30, 150°·s), leg extension power (simultaneous hip and knee extension: 0.8 m·s), squat and countermovement jump height, rebound jump index (jump height·contact time), 20-m linear sprint and change-of-direction (Pro-agility and V-cut tests) time, and aerobic capacity (maximal oxygen uptake, energy cost at LT, and velocity at maximal oxygen uptake and LT) on a level surface. Throughout the training sessions, HR during running did not differ between the groups (DRG: 77.7% ± 4.6% vs LRG: 76.4% ± 4.6% of maximal HR; average across all sessions), while velocity was significantly higher for DRG (14.5 ± 1.1 vs 12.0 ± 1.9 km·h). After the training, DRG significantly improved knee extension torque at all angular velocities (9%-24%) and change-of-direction time for both tests (2%-3%), with no changes in other parameters. LRG significantly improved maximal oxygen uptake (5%), energy cost at LT (3%), and velocity at maximal oxygen uptake (7%), without changes in others. These results suggest that DR training has a greater potential to improve the knee extension strength and change-of-direction ability, but has little effect on the aerobic capacity, compared with HR-matched LR training.