The chaotic and random porosity of typical catalyst particles continues to present a challenge of quantification. An analysis is presented of the variation of diffusion and reaction performance with catalyst porosity architecture. The analysis uses simple 2D pore networks with O(10(3)) pores. The impact of different pore assemblies is assessed for the special case of a bimodal distribution with equal numbers of micro-and macro-pores. A random pore assembly gives higher tortuosity values than the baseline unity tortuosity parallel bundle with identical surface reactivity, diffusivity, pore volume and specific surface. Improvements in particle reactivity are demonstrated for four illustrative porosity architectures-minimum shielding, partial cruciform, fully cruciform and fractal tree. Each of these assembled structures is more reactive than the random network when diffusional resistances intrude. The best reactivity is for the minimum shielding assembly, being some 800% better than random pores at a Thiele modulus of 1. Improved diffusional fluxes through the larger macro-pores, when they inter-penetrate effectively amongst the smaller pores, explains the improvements predicted. Details of the improvements can be discerned from colour-coded pore environmental plots. The practical potential for designing and subsequently fabricating optimal pore structures should be realised by extending to 3D models for porosity. (C) 2007 Elsevier B.V. All rights reserved.