The dynamic ecosystem model LPJ-GUESS includes explicit representation of vegetation dynamics as well as soil biogeochemistry, and has been widely and successfully implemented in predicting vegetation biomass and carbon cycling at different scales. However, the water cycling for each grid cell in the model is only considering the movement between atmosphere, vegetation and soil, ignoring the lateral water movement between grid cells. A previous study has proposed a distributed scheme in LPJ-GUESS incorporating topographic indices to redistribute lateral water movement, and has demonstrated the impacts on ecological functioning and carbon cycling at the Stordalen catchment, northern Sweden. The topographic indices, extracted based on a Digital Elevation Model (DEM), were based on a single flow (SF) algorithm at 50 m resolution, restricting the flow movement to the downslope cell with maximum gradient. In this study we have incorporated the Triangular Form-based Multiple Flow algorithm (TFM) to redistribute lateral water in LPJ-GUESS and analyzed the influences and differences between the two flow algorithms on runoff prediction as well as carbon cycling estimations. The results indicate that the runoff estimated by the TFM algorithm is more realistic than the SF algorithm. Besides, the comparison with observed runoff data demonstrates the monthly runoff estimated using the SF algorithm tends to overestimate the runoff in May and June as well as in the lower flatter peatland region. For the TFM algorithm, the underestimated runoff during the growing season can be compensated by the decreased soil depth in the elevated area. Moreover, the implementation of the TFM algorithm results in a significant increase of the catchment mean value of vegetation uptake of carbon as well as net ecosystem exchange carbon. We conclude that the advanced multiple flow algorithm (TFM) with more accurate estimation of flow accumulation can improve the hydrological predictions in LPJ-GUESS. Meanwhile, the results have proved that the flow routing algorithms do influence the vegetation pattern estimations for the study area.