The mass transfer kinetics of a few compounds (uracil, 112Da), insulin (5.5kDa), lysozyme (13.4kDa), and bovine serum albumin (BSA, 67kDa) in columns packed with several types of spherical particles was investigated under non-retained conditions, in order to eliminate the poorly known contribution of surface diffusion to overall sample diffusivity across the porous particles in RPLC. Diffusivity across particles is then minimum. Based on the porosity of the particles accessible to analytes, it was accurately estimated from the elution times, the internal obstruction factor (using Pismen correlation), and the hindrance diffusion factor (using Renkin correlation). The columns used were packed with fully porous particles 2.5μm Luna-C(18) 100Å, core-shell particles 2.6μm Kinetex-C(18) 100Å, 3.6μm Aeris Widepore-C(18) 200Å, and prototype 2.7μm core-shell particles (made of two concentric porous shells with 100 and 300Å average pore size, respectively), and with 3.3μm non-porous silica particles. The results demonstrate that the porous particle structure and the solid-liquid mass transfer resistance have practically no effect on the column efficiency for small molecules. For them, the column performance depends principally on eddy dispersion (packing homogeneity), to a lesser degree on longitudinal diffusion (effective sample diffusivity along the packed bed), and only slightly on the solid-liquid mass transfer resistance (sample diffusivity across the particle). In contrast, for proteins, this third HETP contribution, hence the porous particle structure, together with eddy dispersion govern the kinetic performance of columns. Mass transfer kinetics of proteins was observed to be fastest for columns packed with core-shell particles having either a large core-to-particle ratio or having a second, external, shell made of a thin porous layer with large mesopores (200-300Å) and a high porosity (≃0.5-0.7). The structure of this external shell seems to speed up the penetration of proteins into the particles. A stochastic model of the penetration of bulky proteins driven by a concentration gradient across an infinitely thin membrane of known porosity and pore size is suggested to explain this mechanism. Yet, under retained conditions, surface diffusion speeds up the mass transfer into the mesopores and levels the kinetic performance of particles built with either one or two porous shells.