The recent accumulation of serially-sampled viral sequences in public databases attests to the need for development of algorithms that infer phylogenetic relationships among such data with the goal of elucidating patterns and processes of viral evolution. Phylogenetic methods are typically applied to contemporaneous taxa, and result in the taxa being placed at the tips or leaves of the tree. In a serial sampling scenario an evolutionary framework may offer a more meaningful alternative in which the rise, persistence, and extinction of different viral lineages is readily observable. Recently, algorithms have been developed to study such data. We evaluate the performance of 5 different methods in correctly inferring ancestor-descendant relationships by using empirical and simulated sequence data. Our results suggest that for inferring ancestor-descendant relationships among serially-sampled taxa, the MinPD program is an accurate and efficient method, and that traditional ML methods, while marginally more accurate, are far less efficient.