This paper is concerned with feedback inhibition of neurotransmitter release by the neurotransmitter itself. We put forward the idea that, similar to multistep biochemical processes, feedback inhibition acts on the initial step in the chain of events that lead to release. Using experimental results carried out on glutamatergic synapses in crayfish, we show that the "first step" hypothesis can account for all experimental results. Our modeling suggests that the biochemical implementation of this inhibition involves the formation of a second messenger, whose production is triggered by binding of transmitter to the autoreceptor. We argue that the autoreceptor is a key part of the release-inducing machinery.