Abstract Rocks from the Chalk Group host important reservoirs of groundwater onshore Denmark and oil and gas in the North Sea. Fine-scale heterogeneity of rocks from the Chalk Group is investigated by stochastic modeling conditioned by first-arrival crosshole GPR traveltimes and lithology data from boreholes. The water-saturated carbonate-dominated rocks contain sharp contrasts among highly porous low-velocity carbonate, thin intercalations of flint, and hardened low-porosity carbonate of higher velocity. The stochastic simulation algorithm can model the contrasting lithologies effectively. It is found that bimodal distributions produce geologically plausible representations of the subsurface. Moreover, the variety of tested stochastic models that honors the specified data uncertainties and prior information provide a good overview of possible subsurface scenarios that the combined data set allows for. Results motivate future GPR-based tracer tests and time-lapse studies for generation of new knowledge of the dynamics of Chalk Group fluid flow.