A very quick decision enables hunting archerfish to secure downed prey even when they are heavily outnumbered by competing other surface-feeding fish. Based exclusively on information that is taken briefly after the onset of prey motion, the fish select a rapid C-start that turns them right towards the later point of catch. Moreover, the C-start, and not later fin strokes, already lends the fish the speed needed to arrive at just the right time. The archerfish predictive C-starts are kinematically not distinguishable from escape C-starts made by the same individual and are among the fastest C-starts known in teleost fish. The start decisions allow the fish-for ballistically falling prey-to respond accurately to any combination of the initial variables of prey movement and for any position and orientation of the responding fish. The start decisions do not show a speed-accuracy tradeoff and their accuracy is buffered against substantial changes of environmental parameters. Here, I introduce key aspects of this high-speed decision that combines speed, complexity, and precision in an unusual way.