We have studied macroscopic current fluctuations associated with the after-hyperpolarization current (I(AHP)) that follows a 200 ms voltage-clamp step to 0 mV in dentate granule (DG) neurones of the rat hippocampus. This maximally effective stimulus produced a peak I(AHP) of 205 ± 20 pA. Background noise was minimized by using the whole-cell single-electrode voltage-clamp configuration. 2. Conventional current-variance analysis was performed on I(AHP) to obtain estimates of the unitary AHP channel current (i) and the maximal attainable AHP current (I(max)). A second approach, utilizing changes in the power spectrum of I(AHP) 'noise' during the decay of I(AHP), was employed to yield an independent estimate of I(max) as well as an estimate of the mean open-state duration of AHP channels. 3. Changes in the power spectrum during I(AHP) decay revealed that the mean channel open time is fixed at 6.9 ± 0.5 ms and that the decay is due to changes in channel closed-state duration. The same analysis gave a value for I(max) of 320 ± 20 pA (n = 7). 4. Current-variance analysis suggests that channels responsible for generation of I(AHP) have a unitary current of 0.29 ± 0.08 pA at -45 mV in 5 mM extracellular potassium and an I(max) of 400 ± 180 (n = 7). Thus, both methods indicate that about 1200 channels are available to generate I(AHP) in DG neurones and that about 60% are open at the peak of a maximal I(AHP). 5. Computer simulations of I(AHP) currents in a model neurone show that dendritic current sources will result in an underestimation of i while I(max) is underestimated to a lesser extent. Estimates of I(max) obtained from power-spectrum analysis are more accurate and less affected by neuronal electrotonic structure than estimates of I(max) based on current-variance analysis.