It was suggested in a recent report by Phillips et al. [J. Neurochem. 43, 479-486 (1984)] that the low-affinity binding of [3H]imipramine in the mouse cerebral cortex could in fact represent binding of [3H]imipramine to the GF/B glass fiber filters used to terminate the assays. The present study demonstrates that this is not the case and advances two lines of evidence: (a) For saturation analysis, mouse cerebrocortical membranes were incubated with [3H]imipramine concentrations between 0.8 nM and 3.6 microM, and parallel incubations were carried out with buffer replacing the brain membranes. The same low-affinity component, in addition to the high-affinity component, was present in the binding of [3H]imipramine to brain membranes plus GF/B filters (uncorrected data), and in that to brain membranes alone (corrected data). (b) Dissociation experiments, in which filter binding is equal for all samples and dissociation time is the only variable, clearly indicated the nonhomogeneity of [3H]imipramine binding. Our results, however, do show that binding to recently purchased GF/B filters is not a negligible phenomenon in saturation experiments. Relatively lower binding was found to GF/C, GF/F, Gelman A/E, and Reeves Angel 934 AH filters; pretreatment of GF/B filters with polyethyleneimine (PEI) reduced binding to a greater extent in the single manifold than in the cell harvester.