The aim of this study was to investigate the role of presumably direct corticospinal pathways in long-term training of the lower limb in humans. It was hypothesized that corticospinal projections are affected in a training-specific manner. To assess specificity, balance training was compared to training of explosive strength of the shank muscles and to a nontraining group. Both trainings comprised 16 1-h sessions within 4 weeks. Before and after training, the maximum rate of force development was monitored to display changes in motor performance. Neurophysiological assessment was performed during rest and two active tasks, each of which was similar to one type of training. Hence, both training groups were tested in a trained and a nontrained task. H-reflexes in soleus (SOL) muscle were tested in order to detect changes at the spinal level. Corticospinal adaptations were assessed by colliding subthreshold transcranial magnetic stimulation to condition the SOL H-reflex. The short-latency facilitation of the conditioned H-reflex was diminished in the trained task and enhanced in the nontrained task. This was observable in the active state only. On a functional level, training increased the rate of force development suggesting that corticospinal projections play a role in adaptation of leg motor control. In conclusion, long-term training of shank muscles affected fast corticospinal projections. The significant interaction of task and training indicates context specificity of training effects. The findings suggest reduced motor cortical influence during the trained task but involvement of direct corticospinal control for new leg motor tasks in humans.