Three different schemes of topography-based runoff production [versatile integrator of surface atmospheric processes (VISA)-TOP1, VISA-TOP2, and VISA-TOP3] are described for a land-surface model (LSM) developed for use with a general circulation model (GCM). The schemes’ sensitivities to some key parameters are assessed for two catchments using data sets developed for the Project for Intercomparison of Land-Surface Parameterization Schemes (PILPS) Phase 2e. VISA-TOP1 differs from VISA-TOP2 only in how to treat oversaturated soil water from the soil layers. In VISA-TOP1, the oversaturated soil water is thrown out of the soil column; hence, it no longer plays a role in the ensuing soil water budgets. In VISA-TOP2, this oversaturated soil water is recharged back to the unsaturated soil layers above the water table; hence, it continues to involve in the water budgets. Unlike VISA-TOP1 and VISA-TOP2, VISA-TOP3 relaxes its dependence on the topographic parameters. The oversaturated soil water is treated the same in both VISA-TOP2 and VISA-TOP3. All three models reproduce the daily and seasonal cycles of streamflow provided that different values of the saturated hydraulic conductivity decay factor are used. The decay factor controls the timing and partitioning of subsurface runoff. In both VISA-TOP1 and VISA-TOP2, an anisotropic parameter explaining different hydraulic conductivities in the vertical and horizontal directions is critical for using the topographic index in the land-surface model. In the VISA-TOP2 scheme, the topography-controlled subsurface runoff is dominant because the oversaturated water is recharged to upper unsaturated soil layers to raise the water table. The water budgets in all these schemes show dramatically different responses to the decay factor, indicating that the calibrated parameters and the model formulations should not be separated. D 2003 Elsevier Science B.V. All rights reserved.