This study was conducted for parametres estimation and stream flow prediction at ungauged catchments on the case of Upper Tekeze basin, Ethiopia by using Rainfall-runoff model. In the basin, most of the catchments were ungauged. The basin has 9199km2 and 3638km2 gauged and ungauged catchments respectively. Rainfall and stream flow data were analyzed in the period of 1992-2006 and 1992-20006, respectively. Parameters calibrated for gauged catchments were extrapolated to ungauged catchments on the base of similar physical catchment characteristics using regionalization techniques. Regionalization methods such as multiple linear regression, spatial proximity, sub basin mean and area ratio were applied to transfer model parameters values from gauged to ungauged catchments. For this study seven gauged catchments were satisfied objective functions in the calibrated and validation period, for example in Gheba catchment Nash-Sutcliffe model efficiency coefficient (NSE), relative volume error (RVE) and coefficient of determination (R2) were, 0.81, -4.25, 0.77 and 0.71, 5.5, 0.74 respectively. Stream flow simulation at ungauged catchments by using spatial proximity and sub basin mean method were contributing high runoff volume compare to other methods. The result for this study shows that the Key model parameters like runoff coefficient (Beta), recession coefficient of upper reservoir zone (Khq), Limit for evapotranspiration (Lp), field capacity (Fc), percolation (Perc) as defaulting value when applying HBV-96 model to the future regionalization studies. Model parameters were calibrated manually by try and error rules, however it was tidies therefore more creative automatic model calibration techniques could be useful for upcoming studies. Thus, Current and future water resources development endeavors may use apply such discharge data for planning and design purposes.