Fault　distribution　type　identification　and　probability　density　function　parameters　estimation　are　the　prerequisite　of　CNC　grinding　machine　tools＇　reliability　assessment.　The　procedure　of　fault　data　grouping,　the　technique　of　probability　density　function　parameters　estimation　based　on　least　square　method　and　the　method　of　reliability　index(mean　time　between　failures,　MTBF)　calculation　are　proposed.　The　probability　density　function　parameters　estimation　and　curve　fitting,　the　comparison　of　goodness　of　fit,　the　MTBF　solution　of　some　type　crankshaft　are　carried　out　by　the　method　proposed　in　the　paper.　The　results　of　statistical　calculation　and　direct　calculation　by　definition　are　close.　The　procedure　of　reliability　assessment　proposed　here　can　be　programmed　easily　and　it　provides　one　measure　for　reliability　assessment　of　CNC　grinding　machine　tools.