This paper presents a framework for application of stochastic discontinuum method for evaluating the uncertainty of hydrocarbon migration around an unlined rock cavern (URC) in water-bearing rock formations. The occurrence of hydrocarbon migration in fractures around the URC was calculated by using a numerical technique based on the migration tracing algorithm and pathway analysis through discrete fracture network (DFN) realizations. The numerical modeling of hydrocarbon migration and uncertainty analysis of fluid flow processes were performed for different arrangement of hydraulic boundary conditions with respect to water curtain pressure head (WCPH) and height of oil in the cavern. The results of numerical simulations demonstrate that the uncertainty and spatial distribution of hydrocarbon migration around URC are very sensitive to the hydraulic boundary conditions, and the geometrical configuration of fractures. The migration probability of both gas and oil decreases non-linearly by increasing the WCPH; however, the rates of decrement for oil and gas are completely different. The maximum migration probability of gas and oil occurred in the case of minimum and maximum height of oil in the cavern, respectively. For most of the hydraulic boundary conditions, the distance and probability of migration for gas are much higher than those for oil. The zones prone to gas and oil migration are mainly concentrated in the neighboring area of the cavern roof and vertical walls, respectively. However, due to the heterogeneous flow system and irregular interconnected fracture networks, the gas and oil migration are possible in the regions below the oil level and below the water bed, respectively. Moreover, the mean and standard deviation of total length and number of migrated fractures around the URC decrease by increasing the WCPH. In addition, both the number of gas migrated fractures and gas escape probability decrease by increasing the height of oil in the cavern. These findings might prove useful for taking optimal decisions in an uncertain framework and for a better design analysis of unlined rock cavern.