Macroscopic, highly disordered, mesoporous materials present a continuing challenge for accurate pore structure characterization. The typical macroscopic variation in local average pore space descriptors means that methods capable of delivering statistically representative characterizations are required. Gas adsorption is a representative but indirect method, normally requiring assumptions about the correct model for data analysis. In this work we present a novel method to both expand the range, and obtain greater accuracy, for the information obtained from the main boundary adsorption isotherms by using a combination of data obtained for two adsorptives, namely nitrogen and argon, both before and after mercury porosimetry. The method makes use of the fact that nitrogen and argon apparently ‘see’ a different pore geometry following mercury entrapment, with argon, relatively, ‘ignoring’ new metal surfaces produced by mercury porosimetry. The new method permits the study of network and pore–pore co-operative effects during adsorption that substantially affect the accuracy of the characteristic parameters, such as modal pore size, obtained for disordered materials. These effects have been explicitly quantified, for a typical sol-gel silica catalyst support material as a case study. The technique allowed the large discrepancies between modal pore sizes obtained from standard gas adsorption and mercury thermoporometry methods to be attributed to the network-based delayed condensation effect, rather than spinodal adsorption. Once the network-based delayed condensation effect had been accounted for, the simple cylindrical pore model and macroscopic thermodynamic Kelvin-Cohan equation were then found sufficient to accurately describe adsorption in the material studied, rather than needing a more complex microscopic theory. Hence, for disordered mesoporous solids, a proper account of inter-pore interactions is more important than that of intra-pore adsorbate density distribution, to obtain accurate pore size distributions.