There　are　two　ways　in　the　development　of　CHF　correlation　in　rod　bundles:　one　is　based　on　round　tube　CHF　correlations,　considering　the　effects　of　spacers　and　CHF　promoters,　effects　of　cold　walls,　etc.　The　other　one　is　to　develop　the　correlation　directly　using　the　rod　bundles　CHF　experimental　data　which　includes　two　methods:　me　minimum　DNBR　point　method　and　the　BO　point　method.　In　this　study,　the　local　thermal-hydraulic　parameters　at　DNB　occurrence　point　of　each　set　of　bundle　CHF　data　were　obtained　by　using　the　subchannel　code　ATHAS.　Both　of　the　minimum　DNBR　point　and　the　BO　point　methods　were　then　applied　to　develop　a　new　CHF　correlation　for　rod　bundles.　The　so-called　＇three-step　method＇　was　used　to　determine　the　shape　of　me　correlation　and　the　non-linear　regression　analysis　method　was　applied　to　determine　the　coefficients　of　the　correlation.　Based　on　the　large　database　of　CHF　tests,　a　new　CHF　correlation　named　the　ACC　correlation　has　been　developed.　The　analysis　and　assessment　results　indicate　that　me　ACC　correlation　can　fit　the　experimental　data　well　with　high　prediction　accuracy　and　correct　parametric　trends.　Compared　with　the　minimum　DNBR　point　method,　me　BO　point　method　is　recommended　to　predict　me　risk　of　DNB　because　it　is　more　reasonable　and　conservative　with　better　predication　rate.　Coupled　with　subchannel　code　ATHAS,　this　correlation　can　simulate　the　thermalhydraulics　performances　of　rod　bundles　exactly.