For interpreting solute transport experiments within rocks, the Residence Time Distributions (RTDs) of solutes in the rock matrix were either derived from the assumption of a homogeneous matrix, or were considered through analytical distributions of diffusion coefficients. A numerical approach based on a Lagrangian framework was developed in order to investigate the effect of spatial heterogeneities on RTDs. The matrix diffusion was simulated over two-dimensional computation grids, representing virtual or real digitized porosity maps. First, virtual porosity maps were used to mimic porous features linked to conductive fractures, such as (i) low-porosity coatings on fracture walls, and (ii) porosity gradients within the rock matrix. Furthermore, RTD was calculated for the real pore network of the Palmottu granite (Finland). It was shown that the arrangement of spatial heterogeneities located in the immediate vicinity of a conductive fracture modifies the RTD of solutes within the rock matrix. Porous zones located near fractures are of particular importance, because they generate anomalies on the RTD.