To explore predator‐prey interactions between California killifish (Fundulus parvipinnis) and spionid polychaetes (Polydora cornuta and Streblospio benedicti) in saltmarsh creeks, we conducted a laboratory flume experiment to quantify whether killifish prey‐patch selectivity varies with water flow. The flume included a 300‐cm2 area of defaunated sediment within which we centrally positioned 24 P. cornuta, 24 S. benedicti, or no worms as a control. We videotaped groups of three killifish for 50 min at one of six unidirectional flow speeds (3 cm s−1, 6 cm s−1, 9 cm s−1, 12 cm s−1, 15 cm s−1, or 18 cm s−1) and recorded their bite rate anywhere in the sediment vs. bites directed at the central patch containing worms (98 cm2). The number of bites anywhere in the 300‐cm2 sediment area increased as flow increased from 3 cm s−1 to 6 cm s−1 and declined linearly as flow increased further. The number of bites anywhere was lower in treatments with no worms than when either worm species was present. The percentage of bites directed at the central patch varied significantly with flow speed and worm presence. With defaunated sediment only, ∼ 33% of bites were directed at the central patch at all flows, consistent with a null model of nonselective foraging. When either worm species inhabited the central patch, ∼ 65% of bites were directed at the central patch at 3 cm s−1 and 6 cm s−1, with patch selection declining as flow increased. Despite differences in morphology and behavior, the two prey species elicited similar foraging activity by killifish. These results suggest the importance of hydrodynamics in driving these epibenthic predator‐prey interactions.