There have been conflicting reports over whether it is possible to stimulate the human cerebellum through the intact scalp using transcranial magnetic stimulation. Here we attempt to clarify the situation in normal subjects by comparing the various methods which have been used. EMG responses evoked by magnetic stimulation over the motor cortex could be suppressed by a prior magnetic stimulus over the cerebellum but the onset latency of the effect varied according to the type of magnetic coil used. Inhibition began at a latency which ranged from 5 to 9 msec in different subjects if conditioning stimuli were given through a flat figure-of-eight coil held horizontally over the basal occiput. The effect lasted a further 6-10 msec. With a larger double cone coil, held vertically over the basal occiput, inhibition began earlier and at a more constant latency of 5 msec. It lasted only 3 msec. Stimulation of the C6/7 nerve roots in the brachial plexus with either an electrical or magnetic stimulus also could suppress EMG responses evoked by cortical stimulation. This began at a conditioning-test interval of 7 or 8 msec and lasted for some 5 msec. We suggest that two types of motor cortical suppression may be elicited from stimulation over the posterior neck/skull: a cerebellar effect starting at 5 msec, and a peripheral nerve effect starting later at 7/8 msec. Stimulation with a horizontal large figure-of-eight coil may produce a mixture of effects because the lower wing of the coil overlaps the posterior neck and can activate peripheral nerve fibres in the brachial plexus.