Counts were made of the number of motoneurons innervating the hind limbs of 10-day normal and paralysed chick embryos whose right hind limb buds had been subjected to varying degrees of amputation prior to innervation. The number of motoneurons on the intact sides of the paralysed embryos was found to be similar to the number present in normal embryos prior to the major period of motoneuron death. Since it has previously been shown that paralysis does not increase the number of motoneurons generated, this means that normal motoneuron death was largely prevented in the paralysed embryos. There were differences in the distributions of motoneurons in the rostrocaudal axis of the spinal cord between normal and paralysed embryos. Therefore, cell death does not eliminate a uniform fraction of motoneurons throughout the rostrocaudal extent of the chick embryo lumbar lateral motor column. It is also argued that there are differences in the relative contribution of the various lumbosacral levels to different parts of the limb, e.g. the shank, before and after the period of cell death. In both normal and paralysed embryos there was a linear relationship between the volume of limb muscle which developed after amputation and the number of motoneurons surviving in the spinal cord. There was no evidence of a 'compression' of motoneurons into the remaining muscle either after amputation alone or after amputation combined with paralysis. Motoneurons are therefore rigidly specified for certain parts of the limb. The relationship between motoneuron number and muscle volume on the amputated side differed from that of the intact side. For a similar increase in muscle volume there was a smaller increase in motoneuron number on the intact sides. This suggested a parallel to the paradoxically small increase in motoneuron number that occurs on the addition of a supernumerary limb.