The strategy decision is the core of enterprise management. Whether or not the enterprise strategy will work, it would directly impact a short-term, or even a long-term, development of an enterprise. The goal of this paper is to establish a new strategic decision model with Bayesian-based approaches, which combine the Bayesian Network with Balanced Scorecard theory. Firstly, starting from the balanced scorecard four dimensions principle, this study constructs the strategic decision model preliminarily. Secondly, it converts the BSC strategic decision map into a Bayesian Network model according to the cause-result relationships existing in both the strategic decision map and Bayesian network. With the study of the Bayesian network structure learning algorithm, we choose the K2 algorithm to optimize the strategy model by deleting some non-critical nodes. Finally, the paper proves the precision and practicability of the model by comparing the BSC index systems with a plurality of specialists' index in various fields.