Motor-evoked potentials (MEPs) after transcranial magnetic stimulation (TMS) show a trial-to-trial variation in size at rest that is positively correlated for muscles of the same, and opposite, upper limbs. To investigate the mechanisms responsible for this we have examined the effect of voluntary activation on the correlated fluctuations of MEP size. In 8 subjects TMS was concurrently applied to the motor cortex of each hemisphere using 2 figure-8 coils. MEPs (n = 50) were recorded from left and right first dorsal interosseous (FDI), abductor digiti minimi (ADM), and extensor digitorum communis. At rest, MEPs were significantly positively correlated for pairs of muscles of the same (75% of comparisons) and opposite limb (56% of comparisons). The correlation for within-limb muscle pairs was strongest for FDI and ADM. In contrast, between-limb MEP correlations showed no somatotopic organization. Voluntary activation reduced the strength of MEP correlations between limbs, even for muscle pairs that remained at rest while a remote upper limb muscle was active. In contrast, activation of a remote muscle did not affect the strength of MEP correlation for muscle pairs within the same limb that remained at rest. For within-limb comparisons, activation of one or both muscles of a pair reduced the strength of the MEP correlation, but to a lesser extent than for between-limb pairs. It is concluded that the process linking corticospinal excitability in the two hemispheres is suppressed during voluntary activation, and that different processes contribute to common fluctuations in MEP size for muscles within the same limb.