Vibration　signals　are　important　information　for　system　failures.　Forecasting　the　trend　of　vibration　signals　is　an　important　content　of　condition　monitoring,　life　estimation　and　fault　diagnosis.　The　advantages,　disadvantages　and　applicable　scope　of　current　trends　and　main　methods　on　vibration　signals　forecasting,　including　time　series　analysis,　grey　prediction,　ANN　(artificial　neural　networks),　SVM　(support　vector　machine)　are　reviewed　in　this　paper.　Difficulties　and　problems　existing　in　vibration　signals　forecasting　and　further　research　directions,　including　nonlinear　and　non-stationary　signal　processing,　combining　forecasting　method,　information　fusion　technology　are　then　put　forward.　Finally,　this　paper　proposes　that　the　above　methods　and　prospects　should　be　developed　into　software　system　for　engineering　application.　(C)　2011　Published　by　Elsevier　Ltd.　Selection　and/or　peer-review　under　responsibility　of　Conference　ESIAT2011　Organization　Committee.