Lifeguard training texts suggest that a lifeguard should continually scan their zone of coverage, carefully examining patrons whose behavior is consistent with drowning or distress. The current study examined whether lifeguard performance is consistent with these specifications, and whether these behaviors have enough visual interest to attract the gaze of non-lifeguards looking for drowning behaviors (“trained”) or those who were given no specified target (“naïve”). Participants viewed video clips of natural swimming taken from three aquatic locations while an eye-tracker recorded their eye position. Lifeguard performance was to some extent consistent with the specifications above, although on many measures it was not statistically better than briefly-trained participants. Implications for future research and training are considered.