Evidence based on electrical studies is presented for the standard and anomalous patterns of innervation of muscles supplied by the fifth lumbar and first sacral nerve roots. Although considerable controversy still exists, previous work in this field suggests that the L5 nerve root supplies tibialis anterior, extensor hallucis longus, extensor digitorum brevis, and the lateral head of gastrocnemius, while the S1 nerve root innervates the medial head of gastrocnemius, soleus, and abductor hallucis. In order to confirm the reliability of this data, the L5 and S1 nerve roots of 50 patients were electrically stimulated during surgery, and distally evoked responses in the relevant muscles were recorded, using surface electrodes. The results confirm the essential reliability of the proposed table of segmental innervation and also demonstrate that most muscles have a dual innervation, with one nerve root being dominant. However, eight patients (16%) exhibited a marked departure from the normal pattern. For example, it is clear that on occasion the extensor digitorum brevis and the lateral head of gastrocnemius can be supplied by S1 and the soleus and medial head of gastrocnemius can be supplied by L5. In a prospective study of 100 patients presenting with clinical evidence of lumbosacral nerve root entrapment, the level of nerve root involvement, as predicted by electromyography, was compared with the operative findings. Correct preoperative nerve root localization was achieved in 84%. At least half the failures in prediction are thought to be due to anomalies of innervation. A further study of 12 patients with disorders of bony segmentation revealed anomalous muscle innervation in seven, so that diagnostic errors may frequently be expected in this group. In clinical practice, whenever anomalous bony segmentation is encountered, the likelihood of variable nerve root innervation should be appreciated, and we recommend that both nerve roots in question should be explored at operation.