The aim of this study was to test the hypothesis that thalidomide acts upon the embryonic peripheral nervous system rather than upon mesenchyme. Pregnant rabbits were given oral thalidomide (150 mg/kg/day) on Days 7-11 of gestation. Fetuses were removed at laparotomy, under anaesthesia, on Day 29 of gestation. Seven fetuses with partial or total absence of the tibia, five treated fetuses without deformities, and four untreated controls were photographed, radiographed, killed and fixed for histological examination. Sciatic nerves were dissected and transverse sections were taken from an identical site. Total fascicular area, myelinated fibre number, fibre density and diameter distribution were obtained. There was a significant reduction in total fascicular area, and in the number of large diameter fibres in all treated animals. There was a significant depletion of total fibre numbers in deformed fetuses compared with controls. These findings are similar to the quantitative changes described in human adult subjects with thalidomide polyneuropathy, and are consistent with primary axonal degeneration in both instances. It is concluded that thalidomide acts upon embryonic nerves rather than on mesenchyme, and that dysmelic deformities of the limbs are secondary to toxic embryonic neuropathy. It is suggested that skeletal defects result when irreversible damage to the nerves reduces the transverse fascicular area below a critical minimum threshold.