Clade dynamics in the fossil record broadly fit expectations from the operation of competition, predation, and mutualism, but data from both modern and ancient systems suggest mismatches across scales and levels. Indirect effects, as when antagonistic or mutualistic interactions restrict geographic range and thereby elevate extinction risk, are probably widespread and may flow in both directions, as when species- or organismic-level factors increase extinction risk or speciation probabilities. Apparent contradictions across scales and levels have been neglected, including (1) the individualistic geographic shifts of species on centennial and millennial timescales versus evidence for fine-tuned coevolutionary relationships; (2) the extensive and dynamic networks of interactions faced by most species versus the evolution of costly enemy-specific defenses and finely attuned mutualisms; and (3) the macroevolutionary lags often seen between the origin and the diversification of a clade or an evolutionary novelty versus the rapid microevolution of advantageous phenotypes and the invasibility of most communities. Resolution of these and other cross-level tensions presumably hinges on how organismic interactions impinge on genetic population structures, geographic ranges, and the persistence of incipient species, but generalizations are not yet possible. Paleontological and neontological data are both incomplete and so the most powerful response to these problems will require novel integrative approaches. Promising research areas include more realistic approaches to modeling and empirical analysis of large-scale diversity dynamics of ostensibly competing clades; spatial and phylogenetic dissections of clades involved in escalatory dynamics (where prey respond evolutionarily to a broad and shifting array of enemies); analyses of the short- versus long-term consequences of mutualistic symbioses; and fuller use of abundant natural experiments on the evolutionary impacts of ecosystem engineers.