Estimation of remaining useful life (RUL) is helpful to manage life cycles of machines and to reduce maintenance cost. Support vectormachine (SVM) is a promising algorithm for estimation of RUL because it can easily process small training sets and multi-dimensionaldata. Many SVM based methods have been proposed to predict RUL of some key components. We did a literature review related toSVM based RUL estimation within a decade. The references reviewed are classified into two categories: improved SVM algorithms andtheir applications to RUL estimation. The latter category can be further divided into two types: one, to predict the condition state in thefuture and then build a relationship between state and RUL; two, to establish a direct relationship between current state and RUL. However,SVM is seldom used to track the degradation process and build an accurate relationship between the current health condition stateand RUL. Based on the above review and summary, this paper points out that the ability to continually improve SVM, and obtain a novelidea for RUL prediction using SVM will be future works.