BACKGROUND Train-of-four stimulation pattern following the administration of non-depolarizing neuromuscular blocking drugs reveals fade on successive contractions. Fade is caused by the release of fewer acetylcholine molecules by the fourth (A4) than by the first stimulus (A1). The current study was conducted to define the relationship between the clinically observed fade and the simulated decline in acetylcholine release (A4/A1) that would be necessary to produce it. METHODS The T4/T1 ratios produced by different doses of vecuronium (15-80 microg x kg-1) were plotted as a function of the concomitant T1. Separately in a model of neuromuscular transmission, T1, T4, and T4/T1 were estimated using simulations in the presence and in the absence of a neuromuscular blocking drug and a stepwise decrease in A4, but constant A1. RESULTS Vecuronium induced neuromuscular block was associated with larger T4/T1 ratios (less fade) during the onset than during the offset of the block. All doses caused similar fade during offset. Simulations revealed that the smallest T4/T1 was associated with the nadir of A4/A1 and occurred at the beginning of T1 recovery. The nadir of A4/A1 was only marginally affected by the dose of vecuronium: 15 microg x kg-1 producing the minimum A4/A1 of 0.8 and 80 microg x kg-1 the minimum A4/A1 of 0.7. CONCLUSIONS The hysteresis in the fade between onset and offset appears to be caused by a delayed decrease of A4/A1 as compared with the decrease in T1. Tentative estimates of the decrease in A4/A1 during fade produced by vecuronium are offered. However, the validity of these estimates is dependent on the validity of the assumptions made in simulations.