Understanding　the　stress-dependent　elastic　moduli　of　fractured　rocks　is　essential　for　monitoring　geopressure　and　tectonic　stress.　The　elastic　nonlinearity　with　finite　strains　can　be　described　by　the　theory　of　acoustoelasticity　through　the　third-order　elastic　constants　(3oECs)　that　are　strictly　valid　for　an　isotropic　homogeneous　medium.　The　extension　to　fractured　rocks　becomes　particularly　interesting　because　of　the　sensitivity　of　fractures　to　prestress,　but　remains　largely　unaddressed.　We　address　this　issue　through　theoretical　modeling　with　confirmed　validity　by　considering　a　group　of　homogeneously　distributed　fractures　(ellipses)　embedded　into　a　homogeneous　background　medium　subject　to　a　uniform　confining　pressure.　We　incorporate　both　the　David-Zimmerman　(DZ)　and　Mori-Tanaka　(MT)　models　into　the　theory　of　acoustoelasticity.　The　DZ　model　accounts　for　the　closure　influence　of　cracks　with　different　aspect　ratios,　whereas　the　MT　model　is　used　to　estimate　the　combining　effect　of　the　background　acoustoelasticity　and　the　elastic　nonlinearity　due　to　the　closure　of　cracks.　We　validate　the　acoustoelastic　DZ-MT　model　of　fractured　rocks　by　experiment　data　with　Fontainebleau　and　Vosges　sandstones.　We　show　that　the　effective　elastic　moduli　of　fractured　rocks　mainly　depend　on　the　background　acoustoelasticity　and　nonlinear　elastic　deformations　induced　by　closing　cracks.　The　inclusions　inside　fractures　have　too　low　3oECs　that　the　associated　acoustoelastic　effect　could　be　neglected　for　both　interconnected　and　isolated　cracks.　The　effective　elastic　moduli　of　rocks　change　significantly　during　the　closing　period　of　cracks　with　increasing　pressure　under　low　pressure.　The　influence　of　cracks　can　be　reduced　only　when　cracks　are　closed　completely,　where,　unlike　the　theoretical　prediction　by　the　DZ　model,　the　effective　elastic　moduli　of　rocks　still　increase　along　with　increasing　pressure　because　of　the　acoustoelastic　effect　of　the　background.