1 The water content and mannitol space of rat diaphragms which were perfused through the inferior vena cava was increased compared with immersed diaphragms. The potassium content of both preparations, when expressed in terms of dry weight, was maintained at similar values to that found in vivo. 2 Despite the application of a constant concentration of decamethonium, a steady level of neuromuscular block was not obtained in either the perfused or immersed rat diaphragm. The immersed preparation differed from the perfused preparation in that recovery from paralysis occurred despite the continued presence of the drug. 3 The rate of uptake of labelled decamethonium (100 muM) at the end-plate region was similar in the perfused and immersed diaphragm. The slopes of the regressions were 0.059 and 0.054 mul mg-1 min-1 (based on dry weight) respectively which were not significantly different. This implies that the rate of uptake of the drug at the end-plate is slow and limited by the rate of entry into the fibre rather than by diffusion to the site of entry.