Integrase(IN)　plays　an　essential　role　in　the　process　of　HIV-1　replication.　IN　inhibitors　of　diketo　acid　derivatives(　DKAs)　were　analysed　by　the　Comparative　Molecular　Field　Analysis(CoMFA)　and　Comparative　Molecular　Similarity　Induces　Analysis(CoMSIA)　methods.　A　set　of　42　compounds　were　randomly　selected　as　the　training　set(35)　and　test　set(7).　Firstly,　a　good　pharmacophore(goodness　of　hit=0.787)　was　obtained　and　used　to　align　ligands.　Then,　predictive　models　were　constructed　with　the　CoMFA　and　CoMSIA　methods　based　on　the　pharmacophore　alignment.　As　a　result,　the　CoMSIA　method　yielded　the　best　model　with　an　r(2)　of　0.955　and　a　q(2)　of　0.665,　which　can　predict　the　activities　of　the　tested　DKAs　very　well(r(2)=0.559).　Finally,　DKAs　were　docked　into　IN,　and　the　predicit　modes　were　superimposed　on　the　contour　maps　obtained　from　the　best　CoMSIA　model.　The　superimposed　maps　gave　a　visualized　and　meaningful　insight　into　the　inhibitory　behaviors,　providing　significantly　useful　information　for　the　rational　drug　design　of　anti-IN　agents.