This paper aims to model the mechanism of knowledge transfer within an inter-firm network. An agent-based simulation model was established to examine the non-linear behavior and the determinant factors of knowledge transfer. Factors such as network size, tie density and link creation rules are included in the model. Eight scenarios were experimented to investigate the impacts of different factors on knowledge transfer performance. The results show that initial network size and link creation rules have significant influence on dynamics of inter-firm network knowledge systems, especially on network structure evolution.