In　geotechnical　engineering,　affected　by　various　geological　processes,　soil　parameters　show　significant　spatial　variability,　and　often　present　non-Gaussian　distributions　and　are　correlated　with　each　other.　Considering　these　characteristics,　it　is　particularly　important　to　introduce　correlated　non-Gaussian　random　fields　in　reliability　analysis.　However,　most　of　the　existing　related　non-Gaussian　random　field　research　has　complex　formulas　and　heavy　calculations.　To　this　end,　this　paper　proposes　a　calculation　method　suitable　for　reliability　analysis　of　rock-soil　collaborative　slopes　based　on　the　strength　reduction　method　and　the　Cholesky　Decomposition　Midpoint(CD-M)　method.　Case　studies　show　that　this　method　can　significantly　improve　the　computational　efficiency　of　slope　reliability　analysis　and　is　suitable　for　reliability　assessment　of　complex　slopes　in　actual　projects.　©　Published　under　licence　by　IOP　Publishing　Ltd.