The number and distribution of acetylcholine (ACh) receptors on muscle cells was studied during development of normal, paralysed and aneural embryonic rat diaphragm muscles. (i) ACh receptors initially are dispersed over the surface of rat embryo myotubes. At day 15 1/2 of gestation junctional receptor clusters (`J-clusters') form in a well ordered band across the midline of the diaphragm muscle; these also form in denervated and paralysed muscles. At about day 18 of gestation additional `EJ-clusters' develop to either side of the midpoint of treated muscles. (ii) If a nerve terminal is present, J-clusters increase in length with time. The time course of generation of new endplates calculated from frequency distributions of J-cluster lengths accurately predicts the muscle growth curve established from muscle fibre counts. (iii) The mean length of J-clusters in paralysed muscles was greater than in controls, due to small new-formed clusters failing to appear. In muscles allowed to recover from paralysis the mean length was less, due to a preponderance of small, new-formed clusters. These observations show that development of new endplates, which is thought to reflect the development of new muscle cells, is halted in paralysed muscles, and recovery from paralysis is associated with the generation of many new endplates. (iv) J-clusters appeared, but failed to grow, in aneural muscles. In muscles denervated during the later stages of gestation, analysis of the distribution of J-cluster lengths shows that new clusters failed to appear, and existing clusters showed little or no increase in length after the time of removal of the nerve. (v) EJ-clusters form by aggregation of dispersed receptors, and their mean length increases with time. They do not appear to be stable entities, and are removed within 2 d of recovery from paralysis. In paralysed muscles, with both J-clusters and EJ-clusters present, only J-clusters attract nerve sprouts or become innervated. (vi) A curve is derived showing development of the total number of synaptic terminals in a muscle. This number increases during days 13-18 of gestation, reaching a peak of about 170% of the adult value during d18 and d19 of gestation. There are two episodes of terminal elimination, one during days 19-21 of gestation, and another about 2 weeks postnatally. During the first postnatal week the number of terminals remains constant at about 140% of the adult number, while the average number of inputs per fibre goes down and the number of muscle fibres increases. (vii) Innervation is essential for muscle development. Motoneurons cannot regulate the number of muscle fibres by requiring a simple one-to-one relation between nerve terminal and muscle fibre, and if their role is regulatory as well as supportive of muscle development then some more complex relationship between nerve terminals and developing myotubes must be postulated.