A stereological analysis was made of synapses in the left and right hemispheres of chick medial hyperstriatum ventrale (MHV) 24h after passive avoidance training (PAL) and in water trained controls (W-control). The synaptic parameters examined were (D), the mean length of the postsynaptic thickening; (NV.syn), the number of synapses per unit volume of neuropil; (VV.syn), the volume density of the pre-synaptic bouton; (NV.ves), the number of synaptic vesicles per unit volume of neuropil and (ves.syn), the mean number of synaptic vesicles per pre-synaptic bouton. No significant differences exist in NV.syn between the left or right hemispheres of W-control and trained chicks, nor is NV.syn influenced by training. However, in W-control chicks D in the right MHV is significantly greater (12%) than in the left MHV and this difference disappears on training. There are no differences in VV.syn between left and right hemispheres of W-control chicks but following training VV.syn is 22.7% greater in the left MHV than in the right MHV. Training als influences the number of synaptic vesicles; in W-control chicks NV.ves in the right MHV is 12.25% greater than in the left MHV but following training these differences are reversed. When the data are expressed as numbers of vesicles per synapse (ves.syn), values for the left hemisphere of trained chicks exceed those in the right hemisphere by a staggering 61.38%. These results are discussed in the context of biochemical and electrophysiological studies which suggest that there is lateralization of the memory trace.