The issue of evaluating equivalent pore diameter distributions in membrane microfilters from gas-liquid (g-l) porosimetry data has been critically examined. Experiments performed with one isotropic and one composite anisotropic membrane in both possible orientations revealed conspicous dependence of the obtained (g-1) porosimetry peaks on imposed pressure ramp rates, rho. Interference of this kinetic effect can be eliminated from the measured data by extrapolation to rho = 0. The ramp rate effect is most likely caused by tortuous pore length distribution, and relatively long times required for liquid expulsion. For two experiments, the observed effects of rho could be reconciled with predictions of the Schlesinger-Bechhold theory [Bechold et al., Kolloid Z., 55 (1931) 172-198]. The data obtained with the thin top layer of the composite membrane facing intruding air directly did deviate somewhat from the theory. Pores characterized by (g-l) porosimetry are likely of the "throat type", and their size distribution is considerably more narrow than that obtained for the "node-type" pores by SEM-image analysis [Zeman and Denault, J. Membrane Sci., 71 (1992) 221-231]. A single bivariate distribution function was constructed for these two distinct pore populations. Flow-weighted or number fraction distributions can be calculated from the extrapolated porosimetry data. For narrow ranges of "throat" diameters, these distributions are fairly similar.