SUMMARY1. The margin of safety for neuromuscular transmission in the tibialis and sartorius muscles of the cat has been determined by measuring the ratio by which end-plate depolarization produced by succinylcholine, decamethonium, octamethonium or iodocholine is antagonized, in the presence of neuromuscular block produced by tubocurarine, gallamine or DF-596. The estimate of the margin of safety was independent of the particular drugs chosen for the measurement.2. To produce threshold block to indirect stimulation once every 10 sec, a fractional occupancy by the antagonist of 0-76 + 0.05 (S.D.) was required; for nearly complete block, an occupancy of 0-917 + 0-16 (S.D.) was required. These figures correspond to factors of safety of 4*1 and 12 for the most sensitive and the most resistant groups of fibres respectively.3. The interaction between the agonists and the antagonists, when tested over a wide range of dosage, did not conform with the conditions of full competitive equilibrium. It was concluded that this arose, not because of some interfering non-competitive process, but because, during the relatively brief exposure to agonist, the equilibrium between the antagonist and the receptors is not significantly disturbed. An analysis of this condition of quasi-equilibrium is given. A correction downwards of the direct estimates of the margin of safety is required, but this proves to be small, about 8 %., and may not be significant.4. The safety factor diminished when the motor nerve had been cut more than 5 hr; it is suggested that this represents an early sign of nerve degeneration.5. With dog sartorius muscle, results similar to those in the cat were obtained. But for deep block in the rabbit, the safety factor was only about 4.6. The existence of a substantial margin of safety influences considerably the interpretation of the time course of action of blocking drugs, and