Impacts of global warming and local scale urbanization on precipitation are evident from observations; hence both must be considered in future projections of urban precipitation. Dynamic regional models at a fine spatial resolution can capture the signature of urbanization on precipitation, however simulations for multiple decades are computationally expensive. In contrast, statistical regional models are computationally inexpensive but incapable of assessing the impacts of urbanization due to the stationary relationship between predictors and predictand. This paper aims to develop a unique modelling framework with a demonstration for Mumbai, India, where future urbanization is projected using a Markov Chain Cellular Automata approach, long term projections with climate change impacts are performed using statistical downscaling and urban impacts are simulated with a dynamic regional model for limited number of years covering different precipitation characteristics. The evaluation of the statistical downscaling methodology over historical time period reveals large underestimation of the extreme rainfall, which is improved effectively by applying another regression model, for extreme days. The limited runs of dynamic downscaling models with different stages of urbanization for Mumbai, India, reveal spatially non uniform changes in precipitation, occurring primarily at the higher quantiles. The statistical and dynamical outputs are further integrated using quantile transformation for precipitation projection in Mumbai during 2050s. The projections show dominant impacts of urbanization compared to those from large scale changing patterns. The uniqueness of this computationally efficient framework lies in an integration of global and local factors for precipitation projections through a conjugal statistical-dynamical approach.