Percutaneous stimulation of human motor cortex electrically (focal anode) and with magnetic coils (MCs) of various designs is compared. The theoretical prediction was confirmed that positioning the standard round MC laterally and orientating it more towards the vertical induces an electric field appropriate for directly exciting corticospinal neurons (cf., the conventional tangential orientation at the vertex). Thus, during voluntary contraction, minimal latency compound motor action potentials (CMAPs) in contralateral arm were elicited both by focal anodic and appropriately orientated MC stimulation. Conduction time from motor cortex to motoneuron was estimated by subtracting peripheral conduction time and monosynaptic delay at the motoneuron from the overall CMAP latency, yielding an estimated corticospinal conduction velocity as high as 66 m/sec. Discontinuous latency variations observed in population CMAPs or individual motor units approximated mono- or polysynaptic cortical synaptic delays and, therefore, are attributed to the intervals between direct and early, or late indirect corticospinal discharges. A TV computer system was used to track movements of individual digits and the hand following MC stimulation. An appropriately orientated MC readily elicited movements predominantly of a single digit, implying focal activation of motor cortex. A double square and a small pointed MC proved especially convenient for eliciting reproducibly single digit movements. Stronger stimulation revealed a topographical gradient in the responses of the different digits. Responses to a given MC stimulus a little above threshold were variable in amplitude, which could not be explained by the relationship of stimulus to phase of the cardiac or respiratory cycle. Overall, our findings indicate the importance of appropriately orientating a standard round MC and using a specially designed MC to obtain the various types of motor response to stimulation of cerebral cortex.