High resolution petrophysical analyses were carried out on Urgonian (Lower Cretaceous) carbonates from outcrops in Provence, SE France. Porosity and permeability were measured on 541 plug samples selected from grain-supported carbonates analogous to those in the age-equivalent Shu'aiba and Kharaib Formation reservoirs in the eastern Arabian Plate. The sampling strategy allowed property heterogeneities from centimetre to kilometre scales to be investigated, as well as correlations between porosity and permeability in several different reservoir rock types. Property spatial modelling sensitivity analyses were also undertaken. The relative abundance of microporosity, grain size and sedimentary-diagenetic anisotropy were the main geological parameters which controlled the petrophysical characteristics of the grainstones studied. Increasing microporosity decreased permeability but resulted in an increase in the homogeneity of the reservoir rocks and therefore in their predictability. An increase in grain size, from fine sand to gravel, and in the amount of intergranular pores, enhanced permeability significantly but resulted in a decrease in the homogeneity (and therefore predictability) of the reservoir rock. At a plug scale, poro-perm relationships are very good and can be used predictively for fine grainstones dominated by microporosity; but relationships are moderate to weak for coarse rudstones with mixed pore types, including intraskeletal pores. In grainstone units, weak sedimentary anisotropy, such as decametre-scale cross-bedding, did not prevent the prediction of the horizontal property distribution from vertical data over a few hundreds of metres. In these units, the lateral correlation of rock properties follows periodic variograms with a 7 m wavelength. The lateral distribution of properties in coarse-grained and heterogeneous rudstones with complex pore types and intense sedimentary heterogeneities, such as channel structures, was however more difficult to predict from a vertical data set. Upscaling poroperm data from plug scale to reservoir scale is linear in the case of grainstones with intergranular microporosity, but is non-linear in the case of skeletal rudstones with coarser pore types including skeletal porosity.