1. Age changes in spontaneous and evoked transmitter release, in receptor number and in ultrastructure at the neuromuscular junction were studied in the CBF-1 mouse strain, which stays physically active and relatively free of organ pathology into advanced age.2. Spontaneous miniature end-plate potentials (m.e.p.p.s) were recorded in the following young (8-12 months) and old (29-33 months) mouse muscles: extensor digitorum longus (e.d.l.), soleus (sol.), gluteus maximus (g.m.), diaphragm (diaph.) and extensor digitorum communis (e.d.c.).3. M.e.p.p. amplitudes were unchanged with age in four muscle groups despite increases in input resistance (in e.d.l., sol. and g.m.). M.e.p.p. amplitude in old diaph. increased 54% with no change in input resistance. Bimodal distributions of m.e.p.p. amplitudes were observed in 6-23% of muscle fibres but were not more prevalent in old mice. There was little or no change in resting membrane potential with age.4. Numbers of junctional acetylcholine receptors (measured with (125)I-alpha-bungarotoxin) were the same in all young and old muscles except e.d.l., where a 30% decrease was noted. Extrajunctional receptors and other indicators of denervation (decreases in resting potential, twitch tension or muscle fibre diameter) were absent or minimal.5. M.e.p.p. frequency decreased in e.d.l., sol. and e.d.c. but not in g.m. or diaph. There was no correlated change in the cholinesterase-positive end-plate area.6. It is concluded that m.e.p.p. amplitude is maintained in old muscles by a combination of compensatory changes. The decline in m.e.p.p. frequency varies between muscle groups and is independent of the length of the motoneurone axon or level of innervation.7. Evoked end-plate potentials (e.p.p.s) were recorded in e.d.l., sol. and diaph. from young (11-13 months) and old (29-30 or 34-35 months) male CBF-1 mice in curarized preparations stimulated at 2 or 20 Hz. The amplitude of the initial e.p.p. of the trains was increased by 122% in old e.d.l. and 93% in old sol., and plateau e.p.p. amplitudes were also increased by about 100% (e.d.l.) and 67% (sol.). This, combined with the absence of change in m.e.p.p. amplitude with age, suggests that the number of quanta released per nerve impulse was increased. In diaph. there was no change with age.8. In all muscle groups, the threshold for initiation of the muscle action potential was unchanged with age. Thus, the relative safety factor of transmission was increased in curarized old e.d.l. and sol. (but not diaph.).9. Depression of the indirect twitch in solutions with a decreased calcium: magnesium ratio was also used as a relative measure of synaptic efficacy. Old sol. and e.d.l. but not diaph. muscles showed less depression of indirect twitch amplitude than did young muscle under these conditions.10. In cut-fibre preparations of sol. and diaph. stimulated at 20 Hz, there was no age-dependent difference in e.p.p. amplitude, in directly measured quantal content, or in curare sensitivity. In view of other results, these findings require careful interpretation.11. Ultrastructural morphometry was carried out in e.d.l. The nerve terminals in old (30 and 34 months) e.d.l. muscles exhibited pronounced loss of synaptic vesicles. In 34-month animals, decreased nerve terminal area and post-synaptic folds devoid of nerve terminals were often observed. Since no evidence of denervation was found by physiological criteria, it is concluded that in 34-month mice, nerve terminals withdraw from some synaptic gutters but do not abandon any junction entirely. The large presynaptic ultrastructural changes contrast with the physiological data showing no deficit and even increases in transmitter release. Therefore, under these conditions, these profound structural changes are either not functionally significant or are well compensated.