This paper presents the results of a streamflow modelling exercise for Slapton Wood catchment using TOPMODEL (Beven & Kirkby, 1979; Beven, et al., 1995) within an uncertainty framework. For the period 1989 to 1991, flow at the main weir was modelled from rainfall measurements and evaporation estimates from the Institute of Hydrology's automatic weather station sited in the catchment. The hydrological record was split into four periods; a calibration period in late 1990 and three validation periods. The Generalised Likelihood Uncertainty Estimation (GLUE) technique of Beven & Binley (1992) was used to estimate uncertainty bounds on flow predictions for the three validation periods, conditioning the predictions on likelihoods derived from the calibration period. Observed flows in the validation periods are enclosed by the uncertainty bounds for most of the time. However, there was some evidence that flow prediction performance was hampered by the effects of structural changes to the model introduced in an attempt to model spatially distributed soil moisture.