Single axon excitatory connections between pairs of neocortical pyramidal neurons were studied using paired intracellular recordings in layers II/III and IV of coronal slices of adult rat somatosensory/motor cortex. Excitatory postsynaptic potentials evoked with different presynaptic firing patterns and at different postsynaptic membrane potentials were compared. Two methods of statistical analysis were used in attempts to determine whether changes in mean excitatory postsynaptic potential amplitude were due to presynaptic or postsynaptic modifications. Analysis of the decrease in mean excitatory postsynaptic potential amplitude associated with increases in presynaptic firing rate were consistent with a change in probability of transmitter release. Paired pulse depression appeared to exhibit both presynaptic and postsynaptic components when the interspike interval was < 10 ms, but could be explained simply by a decrease in probability of release with interspike intervals between 10 and 80 ms. Previous studies had demonstrated that these excitatory postsynaptic potentials are partially mediated by N-methyl-D-aspartate receptors. In contrast to the apparently presynaptic effects of firing pattern, postsynaptic membrane depolarization appeared to produce an increase in quantal amplitude. In addition to this increase at low frequencies, a form of frequency-dependent, self-potentiation involving the recruitment of an additional, longer-latency postsynaptic component occurred at higher presynaptic firing rates. The possibility is discussed that two different mechanisms are involved in the replacement of vesicles at release sites. Over a few tens of milliseconds (paired-pulse depression) availability of releasable transmitter may be determined by the rate of replacement of discharged vesicles from a readily releasable pool of vesicles. Over longer periods of firing at 0.33-2 Hz, the readily releasable pool may become exhausted and require replenishment. Postsynaptic depolarization increases the duration of these excitatory postsynaptic potentials, facilitating summation and enables two components of excitatory postsynaptic potential enhancement at N-methyl-D-aspartate receptor-mediated synapses; one that is present at all firing rates and relates simply to voltage dependent events and one that occurs at higher firing rates and involves a gradual, time dependent event. These data also indicate that the optimal pyramidal firing pattern if another pyramid is to be activated is a tonic, or brief burst pattern at relatively low repetition rates. Long bursts of many presynaptic spikes recruit little that is not activated by pairs of spikes. This situation is in stark contrast to the results obtained in the following paper in which excitatory inputs from pyramids to non-pyramids are described.