In the rat, non-invasive transcranial magnetic stimulation (TMS) has shown promise for evaluation of transmission through the spinal cord before and after repair strategies, but it is still unclear which pathways are activated by TMS. The aim of the present study was therefore to identify these pathways and to analyse the effect of TMS on spinal neurons. In 19 rats, TMS evoked responses bilaterally in forelimb (biceps brachii; BB) and hindlimb muscles (tibialis anterior). The latency and amplitude of these motor-evoked responses (MEPs) were highly variable and depended strongly on the coil position and the stimulation intensity. The most frequently observed latencies for the BB MEPs could be divided into three groups: 3-6 ms, 8-12 ms and 14-18 ms. Lesions in the dorsal columns, which destroyed the corticospinal tract at C2 and C5, significantly depressed MEPs in the mid- and high-latency ranges, but not those in the low-latency range. Lesions in the dorsolateral funiculus, which interrupted the rubrospinal tract, had no effect on MEPs in any of the latency ranges. By contrast, bilateral lesion of the reticulospinal tract and other ventro-laterally located descending pathways abolished all responses. Intracellular recordings from 54 cervical motoneurons in five rats revealed that TMS evoked excitatory postsynaptic potentials (EPSPs) at latencies that corresponded well with those of the BB MEPs. The short-latency EPSPs had rise times of around 1 ms, suggesting that they were mediated by a monosynaptic pathway. EPSPs with longer latencies had considerably longer rise times, which indicated conduction through polysynaptic pathways. Selective electrical stimulation of the pyramidal tract in the brainstem was performed in seven rats, where intracellular recordings from 70 motoneurons revealed that the earliest EPSPs and MEPs evoked by TMS were not mediated by the corticospinal tract, but by other descending motor pathways. Together, these results showed that in the rat TMS activates several descending pathways that converge on common spinal interneurons and motoneurons. Our observations confirm that the corticospinal tract has weak (and indirect) projections to cervical spinal motoneurons.