1. Spontaneous and evoked transmitter release at the crayfish neuromuscular junction were potentiated in response to photolytic release of calcium from the 'caged' calcium compound DM-nitrophen, which had previously been injected into presynaptic terminals. 2. The amount of calcium released from DM-nitrophen photolysis depends on the concentration of DM-nitrophen, its photoproducts, Ca2+, Mg2+, H+, ATP and the cell's native buffer. Since none of these are known in the crayfish terminal, the study was conducted in a qualitative fashion. 3. Photolytic release of calcium from DM-nitrophen increased excitatory junctional potentials (EJPs) by a range of 2-31 times over control values and the miniature excitatory junctional potential (MEJP) frequency increased from resting values of 1-10 quanta/s to 3000-11,000 quanta/s. 4. Extracellular calcium was not required for the light-evoked asynchronous release of transmitter. Calcium-bound DM-nitrophen previously pressure injected into crayfish presynaptic terminals increased the MEJP frequency from resting values of 1-8 quanta/s to 800-10,000 quanta/s during photolysis in a calcium-free cobalt Ringer solution. 5. Iontophoresis of calcium-free DM-nitrophen into presynaptic terminals released transmitter upon photolysis, but only in a calcium-containing Ringer solution. This suggests that DM-nitrophen is capable of binding calcium once injected into terminals, but this is dependent on the presence of external calcium. 6. Photolysis of DM-nitrophen at lower light intensities produced a slower rate of transmitter release. 7. Brief light exposures, i.e. those which photolysed 5-20% of the DM-nitrophen, resulted in a rapid decay of postsynaptic responses on extinguishing the light, due to rebinding of photolytically released calcium to unphotolysed DM-nitrophen. Longer light exposures which completely photolysed DM-nitrophen, leaving only the low affinity photoproducts, produced a slow decay of transmitter release after the light pulse, presumably due to the active extrusion of calcium from the presynaptic terminals. 8. During photolysis of DM-nitrophen, the time courses of changes in EJP amplitude and MEJP frequency were different, indicating that the two measures of transmitter release were not linearly related. 9. MEJP frequency and EJP amplitudes during DM-nitrophen photolysis were fitted to a 'non-linear summation model' in which photolytically released calcium sums with calcium entering during an action potential to evoke transmitter release with a calcium co-operativity of five.