Advances in quantitative methodologies and emerging new big geo-data have provided great opportunities for quantifying indicators and modelling urban phenomena at the micro level. However, there are still many challenges in modelling urban phenomena at the micro lev-el: e.g., high-resolution data requirement, long computing times, spatial representation, the extent to which the model can be replicated, the way the model can be communicated and shared with others. In this paper we aim to address the challenge of spatial representation in ur-ban modelling at the micro level by advocating an idea of emphasizing the significance of using street units in urban modelling. We reviewed the typical urban models and discussed why urban modelling focusing on street should be advocated. Then we proposed a framework of ur-ban modelling on streets and gave several studies within the proposed framework for illustration. We concluded that urban modelling on streets are necessary and promising to further support planning- and policy- making.