While multiple hypothesis tracking (MHT) is widely acknowledged as an effective methodology for multi-target surveillance, there is a challenge to manage effectively a potentially large number of track hypotheses. Advanced single-stage track-while-fuse does not always offer the best processing scheme. We study two instances where multi-stage MHT processing is beneficial–dense target scenarios and complementary-sensor surveillance–and propose two processing schemes for these challenges: track-break-fuse and track-before-fuse, respectively. We provide simulation results demonstrating the advantages of these schemes over track-while-fuse. More generally, we argue that multi-stage MHT offers a powerful and flexible paradigm to circumvent limitations in conventional MHT processing.