The extent of liquid back-mixing in gas-liquid concurrent upflow packed-bubble column reactors is quantified in terms of an axial dispersion coefficient or its corresponding dimensionless Peclet number. Effects of reactor operating conditions on the axial dispersion coefficient are not properly accounted for by the available literature correlations, wherein most often the Peclet number is expressed solely in terms of the gas and liquid Reynolds numbers or superficial velocities. Based on the broadest experimental databank (1322 measurements, II liquids, four gases, 28 packing materials, 14 columns diameters, Newtonian, non-Newtonian, aqueous, organic, coalescing and non-coalescing liquids, high pressure, bubble and pulsing how regime conditions), a state-of-the-art liquid axial dispersion coefficient correlation is obtained by combining neural network modeling and dimensional analysis. Thorough qualitative and quantitative analyses of the constructed databank demonstrate the robustness of the proposed correlation to restore the variety of trend variations of liquid Peclet numbers reported in the literature. (C) 2001 Elsevier Science B.V. All rights reserved.