The study proposes a verification mechanism based on reachability analysis of pushdown system to enforce existing declassification policies of language-based information flow security. The pushdown rules of store and match primitives are embedded in the abstract model after compact self-composition. The security property with respect to different declassification policies is violated when the illegal-flow state is reached in the pushdown system. The experimental results show improvement in precision, compared with the type-based mechanisms, and growth in effectiveness compared with the RNI-enforcement based on automated verification.