The motor cortex contains a distributed map of muscles, with a single muscle represented over a wide cortical area. We have searched for inter-connections between distant sites projecting to common muscles by delivering pairs of 20-microA single-pulse intracortical microstimuli (ICMS) to sites separated by 1.5-2 mm in the hand-area primary motor cortex of two macaque monkeys performing a precision grip task. The facilitation of hand- and forearm-muscle rectified EMG was measured. When stimuli were delivered simultaneously, responses were quantified using a technique to correct for non-linearities inherent in the use of averaged, rectified EMG. A spatial facilitation was seen for such simultaneous stimuli; however, it was of the same magnitude as that occurring when ICMS was paired with stimulation of corticospinal axons in the pyramidal tract (PT), so that it was likely to be spinal in origin. When two such distant sites were stimulated separated by a 10- or 20-ms delay, the second response scaled with the level of background EMG in the same way as a response to the PT stimulus. By contrast, when the same site was stimulated twice with these delays, the second response showed a facilitation compared with a similarly timed PT response. There would therefore appear to be a local facilitation of the cortical output at these intervals, which is not seen between distant sites. Antidromically identified pyramidal-tract neurones (PTNs) were recorded whilst stimuli were delivered to a cortical site, with a distance between stimulating and recording electrodes of also 1.5-2 mm. The most common response was a facilitation followed by a suppression. Six of eleven PTNs showed a facilitation in their discharge following this stimulation (maximum connection strength s=0.19), 8/11 showed a suppression (maximum s=0.16). It is concluded that powerful inter-connections do exist between distributed parts of the motor output and that there is widespread cortical activation after even a single ICMS pulse. However, these inter-connections do not lead to interactions between cortical outputs following stimulation, as assessed from the EMG. It is proposed that this is likely to reflect differences in the summation of output cells to local versus remote stimulation.