Paired-pulse field responses were recorded from the granule cell layer of the dentate gyrus in brain slices from temporal lobe epileptic patients. Paired-pulse depression (PPD) was examined using perforant path stimulation of low to moderate intensity at an inter-stimulus interval (ISI) of 20 ms. The paired-pulse ratio (PS2/PS1) was expressed as the population spike amplitude of the second response (PS2) relative to that of the first response (PS1). Representative tissue response from each patient biopsy were divided into two groups that were significantly different based on the magnitude of the highest paired-pulse ratio recorded for each biopsy specimen: the strong paired-pulse depression group (PS2/PS1 = 0.12 +/- 0.03; n = 15) and the weak paired-pulse depression group (PS2/PS1 = 0.68 +/- 0.06; n = 13). Paired-pulse ratios from the strong PPD group were relatively independent of stimulus intensity, whereas, PPD was dependent on stimulus intensity in the weak PPD group; i.e., PPD was greatest at the lowest intensity and reached a plateau at higher intensities. Bicuculline (20 microM) and low concentrations of baclofen (0.1-0.2 microM) reduced paired-pulse depression in the strong PPD group, but did not significantly change the paired-pulse ratio in the weak PPD group. Paired-pulse facilitation was observed in some cases after inhibition was blocked pharmacologically. The number of population spikes was increased in the presence of bicuculline but was unchanged by baclofen. In the strong PPD group, baclofen significantly altered the EPSP-population spike (E-S) relationship by increasing the slope of the relationship for the second response, without having an effect on the slope of the first response. Baclofen had no effect on the E-S relationship of either response in the weak PPD group. The data are consistent with (1) less inhibition in the weak PPD group compared to the strong PPD group, (2) reduction of feedback inhibition in the strong PPD group by bicuculline and by low concentrations of baclofen, and (3) the occurrence of paired-pulse facilitation when inhibition was pharmacologically reduced in the dentate gyrus of temporal lobe epileptic patients. The results are also consistent with the presence of GABAB receptors on human inhibitory interneurons that, when activated by baclofen, result in disinhibition of granule cells through feedback circuits. Although inhibition may be compromised in some epileptic human biopsy specimens, the presence of strong inhibition in other patients' biopsy material suggest the re-evaluation of the role of inhibition in epilepsy.