The aim of the study was to establish the effects of human-induced watershed changes on strearnflows. The research hypothesised that land use change influences base flows. Enjoro river in Kenya was used as the case study. In the 1940s, the watershed was characterised with a sparse population, forestry and large scale conservative agriculture. The river regime was naturally perennial. Between 1960 and 1990 land subdivision, intensive cultivation, urbanisation, and deforestation changed such stable ecosystem Several approaches were used to evaluate the perceived cause-effect relations in the watershed. The time series of the flows, rainfall, and other climatic records were used to infer effects of changed physical characteristics in the watershed. Quantitative evaluation of the changes was accomplished by simple graphs, homogeneity tests, satellite imagery and model simulations of hydrologic variables. Analysis of the data series before and after the presumed changes provided an understanding of the variability masked in the hydrologic system. These comparisons allowed for the determination of the period in which the watershed changes influenced the river regime. The combinede ffect of humana nd natural factors decreasedth e river basef lows. A 30% increase in deforestation, 20% in agriculture and 10.4% in urbanisation was observed- Water availability decreased from a runoff coefficient of 22% in the 1960s, 10% in the 1970s and 8% in the 1980s. This progressive decline in runoff developed into hydrologic drought regime in the 1980s. Normalized difference vegetation index (NDVI) predicted well the flow changes in the watershed. Simulations of rainfall and flow supported the changes observed in the hydrologic variables. The optimised parameters with HYRROM showed 'store' parameters (SS, RDEL, GDEL) to be sensitive to changes in vegetation cover especially during the dry years of 1965,1973 and 1984. The model simulated some parameters in the watershed which could be used to infer changes in strearnflows due changes in land use. It was however, difficult to estimate and to validate long-term model parameters because of limited data and the contrasting geography of the region which induced hydrologic variability. The model did not isolate effects of specific land uses, although it predicted the observed flows. There is evidently, a need for future research on the problem. The investigation demonstrated the difficulty in identifying differences in strearnflows from watersheds undergoing simultaneous physical changes and human interventions. Since a specific effect of a particular land use change could not be isolated independently, continued research on the development of an integrating watershed coefficient is recommended. Remote sensing techniques should be incorporated in the development of integrating watershed coefficients.