We conducted a displacement experiment to determine the influence of sensory cues on homing of pelagic fish to fish aggregation devices (FADs). If pelagic fish use chemical cues to home to a FAD, and cues such as odours from fouling organisms and fish associated with the FAD only occur down-current, we hypothesised that fish released down-current from a FAD would return in higher proportions than fish released across-current. Coryphaena hippurus (n = 161) and Seriola lalandi (n = 31) were caught from January to May in 2000, 2001 and 2002 within 50 m of a FAD and displaced to release sites 20, 75, 275 or 500 m across- or down-current. Overall proportions of return of C. hippurus to FADs varied among the 3 yr, with returns highest in 2000 (48.3%), lowest in 2001 (7.6%) and intermediate in 2002 (33%). There was a strong relationship between abundance of C. hippurus around FADs and current speeds, possibly due to a behavioural shift towards closer association with FADs during strong currents. This may have affected the proportions of return in different years, as currents were strongest in 2000 (0.46 m s-1 ± 0.06), weakest in 2001 (0.13 m s-1 ± 0.02) and intermediate in 2002 (0.30 m s-1 ± 0.09). Both C. hippurus and S. lalandi returned in proportions significantly higher than could be expected if fish swam in a random direction from the point of release, indicating a directed homing behaviour. Returns from down- and across-current release sites were similar at all distances for C. hippurus and S. lalandi, providing no evidence to suggest that these species used only chemical cues to home to FADs. As fish returned from up to 275 m away, sensory processes other than vision and olfaction must operate. Sound or vibrations from associated fish and the FAD are possible cues for homing.