Harmful algal blooms (HABs) represent a diverse range of phenomena that universally share only two characteristics: they produce effects on ecosystems or food resources that humans perceive as harmful, and their progression is fundamentally a process of population dynamics under oceanographic control. Because of the complexity, scales, and transient nature of HABs, their monitoring and prediction requires rapid, intensive, extensive, and sustained observations at sea. These requirements cannot be met with traditional approaches that depend on ships for sampling and laboratories for chemical or biological analyses. Fortunately, new sensing technologies that operate autonomously in situ will allow, in the near future, the development of comprehensive observation strategies for timely detection of HABs. In turn, developments in modeling will support prediction of these phenomena, based directly on real-time measurements.