In this paper the constrained project portfolio optimization problem is considered. It is modeled as a parameterized optimization problem, while considering risk, dependency, impact of delay and some other factors which will affect the result for the project portfolio optimization. A heuristic based on dynamic programming and graph theory is proposed to solve the model, and a tool is built to illustrated the validity and efficiency of the model and heuristics by using a real case. The proposed parameterized model and approaches is proved to be valid and efficient. © 2012 IEEE.