The phenomenon of long-term potentiation is widely used as an experimental model of memory. An approach that has been used to study its underlying mechanisms is to analyse its interaction with presynaptic paired-pulse facilitation. Several studies found no evidence for an interaction in the CA1 hippocampal area, whereas other data, for example from quantal analysis, suggested that presynaptic mechanisms contribute to the maintenance of long-term potentiation. In the present study, initial slopes of field potentials in area CA1 were measured in rat hippocampal slices. "Conventional" long-term potentiation was induced by high-frequency (100 Hz) afferent tetanization of the testing input. "Associative" long-term potentiation was induced by combining lower frequency (40 Hz) tetanization of a testing input with high-frequency tetanization of a second input. The paired-pulse facilitation ratio decreased in the majority of experiments in which long-term potentiation was induced conventionally, but it decreased, increased or did not change after inducing associative potentiation. Decreases in the paired-pulse facilitation correlated inversely with the initial (pre-tetanic) facilitation ratio. A more detailed regression analysis suggests that this correlation results from two other correlations: (i) that between changes in paired-pulse facilitation and the magnitude of long-term potentiation, and (ii) that between initial paired-pulse facilitation and the magnitude of long-term potentiation. The first correlation prevailed during the initial 10 min following tetanization, while the second prevailed 40-60 min later. A post-tetanic decrease in paired-pulse facilitation is evidence for an involvement of presynaptic mechanisms in the maintenance of long-term potentiation. The lack of significant changes in some studies could be due to the inclusion in the analyses of experiments with long-term potentiation of small magnitude, in which changes in paired-pulse facilitation ratios would have been inconsistent. The present study suggests that the early (10-20 min) and late (40-50 min) phases of long-term potentiation were mediated by different mechanisms, with a mixture of these mechanisms during the intermediate period. On the basis of the present and previous studies, the following scheme of involvement of several mechanisms in long-term potentiation maintenance is proposed. The early phase includes two major mechanisms: an increase in the probability of transmitter release, leading to an apparent increase in the number of effective release sites, and an increase in efficacy of one transmitter quantum, probably due to an increased number of postsynaptic receptors. The later phase of long-term potentiation is attributed to an increase in the number of transmitter zones, presumably due to structural modifications.