In　this　paper,　we　present　CacBDD,　a　new　efficient　BDD　(Binary　Decision　Diagrams)　package.　It　implements　a　dynamic　cache　management　algorithm,　which　takes　account　of　the　hit-rate　of　computed　table　and　available　memory.　Experiments　on　the　BDD　benchmarks　of　both　combinational　circuits　and　model　checking　show　that　CacBDD　is　more　efficient　compared　with　the　state-of-the-art　BDD　package　CUDD.　©　2013　Springer-Verlag.