When a neurotransmitter substance is released into a synaptic cleft it acts upon subsynaptic receptors to induce a response of the target cell and also interacts with systems which act to remove the substance. At no time is there an equilibrium, and it is inappropriate to apply equilibrium kinetics to predict the consequences of modifying the system, e.g. by blocking receptors. A mathematical model predicts that the subsynaptic response to each package or "quantum" of transmitter may be insensitive to competitive receptor blockade, or to quite large changes in receptor density, provided the density of receptors is normally enough for efficient capture of transmitter. This prediction is borne out by experimental data from the voltage-clamped mouse neuromuscular junction; it requires blockade or removal of about 80% of receptors (90% after poisoning of acetylcholinesterase) to reduce the miniature end-plate current, i.e. the action of a quantum if nerve-released acetylcholine (ACh), by 50%. On the other hand, drugs that interfere with receptor function without preventing ACh binding to receptors can be just as (or more) effective in blocking nerve-applied as in blocking exogenously applied transmitter substances. At the neuromuscular junction this is seen with receptor desensitization, and "non-specific" agents such as local and general anaesthetics. We conclude that care must be taken in extrapolating from data re receptor number and/or occupancy by blocking drugs to consequences in terms of synaptic function.