The Meuse river basin covers an area of 33,000 km2, touches five countries and is a major communication route in Europe. It is one of the catchments with longest streamflow records, with daily measures of discharge dating back to the beginning of the previous century. Attempts to model streamflow with standard hydrological models revealed that average streamflow was consistently overestimated by the model in the period 1933-1968. Different attempts to explain such anomaly can be found in the literature. In this work we hypothesise that this anomaly could be resolved by considering a time varying root zone storage capacity, represented by a model parameter (Su,max), which has affected the partitioning between precipitation and streamflow. Vegetation is in fact believed to adjust root zone storage capacity to overcome droughts with a return period of about 20 years. To test our hypothesis, a semi-distributed conceptual model, based on the FLEX modelling approach, was used. A time varying Su,max was obtained with two approaches: by calibration of the model parameters in a moving time window, and by derivation of Su,max directly from climate variables. The results show that adding time dependency to Su,max improves the mean flow simulation, however not to a degree that it fully explains the observed anomaly. Deriving Su,max directly from climate variables delivered a better fit to the average streamflow than calibration, which confirms the feasibility of a climate derived root zone storage capacity in hydrological modelling.