Migration independently evolved numerous times in animals, with a myriad of ecological and evolutionary implications. In fishes, perhaps the most extreme form of migration is diadromy, the migration between marine and freshwater environments. Key and longstanding questions are: how many times has diadromy evolved in fishes, how frequently do diadromous clades give rise to non-diadromous species, and does diadromy influence lineage diversification rates? Many diadromous fishes have large geographic ranges with constituent populations that use isolated freshwater habitats. This may limit gene flow among some populations, increasing the likelihood of speciation in diadromous lineages relative to non-diadromous lineages. Alternatively, diadromy may reduce lineage diversification rates if migration is associated with enhanced dispersal capacity that facilitates gene flow within and between populations. Clupeiformes (herrings, sardines, shads and anchovies) is a model clade for testing hypotheses about the evolution of diadromy because it includes an exceptionally high proportion of diadromous species and several independent evolutionary origins of diadromy. However, relationships among major clupeiform lineages remain unresolved and existing phylogenies sparsely sampled diadromous species, limiting the resolution of phylogenetically-informed statistical analyses. We assembled a phylogenomic dataset and used multi-species coalescent and concatenation-based approaches to generate the most comprehensive, highly-resolved clupeiform phylogeny to date, clarifying associations among several major clades and identifying recalcitrant relationships needing further examination. We determined that variation in rates of sequence evolution (heterotachy) and base-composition (non-stationarity) had little impact on our results. Using this phylogeny, we characterized evolutionary patterns of diadromy and tested for differences in lineage diversification rates between diadromous, marine, and freshwater lineages. We identified thirteen transitions to diadromy, all during the Cenozoic Era (10 origins of anadromy, two origins of catadromy, and one origin of amphidromy), and seven losses of diadromy. Two diadromous lineages rapidly generated non-diadromous species, demonstrating that diadromy is not an evolutionary dead-end. We discovered considerably faster transition rates out of diadromy than to diadromy. The largest lineage diversification rate increase in Clupeiformes was associated with a transition to diadromy, but we uncovered little statistical support for categorically faster lineage diversification rates in diadromous versus non-diadromous fishes. We propose that diadromy may increase the potential for accelerated lineage diversification, particularly in species that migrate long distances. However, this potential may only be realized in certain biogeographic contexts, such as when diadromy allows access to ecosystems in which there is limited competition from incumbent species.