To search for ultrastructural correlates of differences in synaptic safety factor and neurotransmitter release, neuromuscular junctions from the cutaneous pectoris and cutaneous dorsi muscles of the grass frog Rana pipiens were freeze fractured. Synaptic efficacy in these muscles was determined by the extent to which isometric twitch tension could be blocked by lowering [Ca2+] in the bathing solution. We found that junctions in the cutaneous pectoris were significantly more effective than those of the cutaneous dorsi. Morphometric analysis of 16 junctions from each type of muscle showed significant differences in some aspects of active zone structure. Cutaneous pectoris terminals had longer active zone segments and active zones spaced more closely together. This resulted in 20% more active zone length per unit terminal length in the cutaneous pectoris. Cutaneous dorsi terminals had active zones that were more often segmented into two or more sections at a single junctional fold. Mean active zone length per junctional fold and the number of active zone particles per micrometre of active zone length were not significantly different. As a result of the somewhat larger terminal width in the cutaneous dorsi, the percentage of terminal width occupied by active zone was greater in the cutaneous pectoris. As an attempt to indirectly estimate active zone spacing with the light microscope, we applied rhodamine-conjugated alpha bungarotoxin to neuromuscular junctions from the cutaneous pectoris and cutaneous dorsi. No significant difference in the spacing of fluorescently labelled acetylcholine receptor bands was found between the two types of junctions. Our results indicated that the greater active zone length per unit terminal length in the cutaneous pectoris was associated with its increased synaptic efficacy. In addition the continuity and particle organization of active zones may have contributed to the observed differences in synaptic safety factor at frog neuromuscular junctions.