Multiple hypothesis tracking (MHT) addresses difficult association problems in multiple target tracking by forming and evaluating data association hypotheses with multiple scans or frames of data. This paper reviews 40 years of MHT research and development since publication of the measurement-oriented MHT journal paper in 1979. It covers hypothesis-oriented and track-oriented MHT, distributed MHT, graph-based association, other MHT research, and the relationship with multitarget filters using random finite sets. It also reviews use of MHT in surveillance and other applications.