In normal subjects, focal repetitive transcranial magnetic stimulation (rTMS) of the hand motor area evokes muscle potentials (MEPs) from muscles in the hand (target muscles) and the arm (non-target muscles). In this study we investigated the mechanisms underlying the spread of MEPs induced by focal rTMS in non-target muscles. rTMS was delivered with a Magstim stimulator and a figure-of-eight coil placed over the first dorsal interosseus (FDI) motor area of the left hemisphere. Trains of 10 stimuli were given at a suprathreshold intensity (120% of motor threshold) and at frequencies of 5, 10 and 20 Hz at rest. Electromyographic (EMG) activity was recorded simultaneously from the FDI (target muscle) and the contralateral biceps muscle and from the FDI muscle ipsilateral to the side of stimulation (non-target muscle). rTMS delivered in trains to the FDI motor area of the left hemisphere elicited MEPs in the contralateral FDI (target muscle) that gradually increased in amplitude over the course of the train. Focal rTMS trains also induced MEPs in the contralateral biceps (non-target muscle) but did so only after the second or third stimulus; like target-muscle MEPs, in non-target muscle MEPs progressively increased in amplitude during the train. At no frequency did rTMS elicit MEPs in the FDI muscle ipsilateral to the site of stimulation. rTMS left the latency of EMG responses in the FDI and biceps muscles unchanged during the trains of stimuli. The latency of biceps MEPs was longer after rTMS than after a single TMS pulse. In conditioning-test experiments designed to investigate the cortical origin of the spread, a single TMS pulse delivered over the left hemisphere at an interstimulus interval (ISI) of 50, 100 and 150 ms reduced the amplitude of the test MEP evoked by a single TMS pulse delivered over the right hemisphere; and a conditioning rTMS train delivered over the left hemisphere increased the amplitude of the test MEP evoked by a single TMS pulse over the right hemisphere. A conditioning rTMS train delivered over the left hemisphere and paired magnetic shocks (test stimulus) at 3 and 13 ms ISIs over the right hemisphere reduced MEP inhibition at the 3-ms ISI but left the MEP facilitation at 13 ms unchanged. Using a control MEP size matched with that observed after a conditioning contralateral rTMS, we found that paired-pulse inhibition remained unchanged. Yet a single TMS conditioning pulse sufficiently strong to evoke a MEP in the contralateral FDI and biceps muscles simultaneously (as rTMS did) left paired-pulse inhibition unchanged. We conclude that the spread of EMG activity to non-target muscles depends on cortical mechanisms, mainly including changes in the excitability of the interneurones mediating intracortical inhibition.