A fuzzy-parameterised stochastic modelling system (FPSMS) was proposed in this study to investigate the impacts of uncertainties associated with hydrocarbon contaminant transport in subsurface. FPSMS integrated the multiphase numerical simulator, the fuzzy transformation method, and the Monte Carlo simulation technique into a general modelling framework, and was capable of dealing with coupled probabilistic-possibilistic uncertainties (in fuzzy-parameterised stochastic format). The simulation of light non-aqueous phase spill liquid (LNAPL) in an experimental system was used to demonstrate the applicability of the proposed method. Porosity and intrinsic permeability were considered as stochastic inputs with the means and standard deviations being characterised by fuzzy sets. The study results demonstrated that FPSMS was effective in evaluating the joint impacts of highly uncertain inputs on predictions of the LNAPL movements in subsurface. Compared with traditional fuzzy-stochastic analysis methods, FPSMS was suitable in tackling dual uncertainties, generating outputs with richer information, and even having more efficient calculation algorithms. Also, it could be a good reference for further risk assessment and remediation design for petroleum-contaminated sites.