As expected global increase in stream temperature triggered by climate change would alter benthic community composition of streams, it is necessary to understand the future variation of benthic invertevrates. First, we developed simple linear regression model to estimate the density of invertebrates by a water temperature metric as predictor variable. These data for the model were monthly collected during May-October within headwater streams which were homogenous in hydro-morphological characteristics. Secondly, we projected water temperature distribution within this basin for the two period in future using a distributed hydro-thermal simulation model which needed GCM data as inputs. Finally, we projected the variation of the density of benthic assemblage based on the regression model by annual mean of water temperature.