The groundwater flow in karst aquifers is prone to become non-Darcian due to the presence of karst conduits, and this effect may become more pronounced during tunnel construction. Prediction of discharge into tunnels in karst aquifers remains a challenging task because of the complex nature of non-Darcian flow and the uncertainty of aquifer's properties. This study aims to evaluate the effect of non-Darcian flow on the discharge into a tunnel in karst aquifers in Guizhou Province, China, where a large proportion (~86%) of borehole packer tests evidences the flow in non-Darcian regime. An integrated method combining packer test data interpretation, statistical correlation and inverse modelling is proposed to determine the non-Darcian flow properties of the aquifer system. Numerical simulations of steady-state flow show that the predicted discharge based on the Forchheimer's equation agrees better with measurements than that based on the Darcy's law by ~27%. The region with the most pronounced non-Darcian effect occurs at the intersection between the tunnel and a karstified fault connecting to an underground river, and the occurrence of non-Darcian flow leads to an increase in hydraulic gradient, but a reduction in discharge into the tunnel. This work presents a framework that combines aquifer properties estimation and numerical simulations for more accurately evaluating the discharge and seepage erosion risk induced by tunnel construction in karst aquifers.