Foraging behavior of Harlequin Ducks (Histrionicus histrionicus) and its response to changing environmental conditions during winter was investigated at Cape St. Mary's, Newfoundland. Behavioral synchrony among individuals permitted continuous observations of flocks to be conducted, sometimes over entire days. Overall foraging effort of Harlequin Ducks was examined at two organizational levels: dive cycles (dive and surface pause) and foraging cycles (foraging bout and rest bout). Overall foraging effort decreased at greater tide depths as Harlequin Ducks decreased the duration of foraging bouts. Dive:pause ratios did not change within these shorter foraging bouts at high tides, however the duration of dives, pauses, and the total dive cycle all increased. Overall foraging effort decreased in response to increased wind/wave exposure due to a decrease in dive duration. Overall foraging effort did not change in response to decreasing ambient temperature; however the frequency of dive cycles decreased which could decrease energy expenditure associated with post-dive thermoregulatory costs. Overall foraging effort increased throughout the day, particularly in the last foraging bout before the overnight fasting period. Interestingly, this strategy was accomplished by decreasing dive durations but increasing foraging bout duration. These opposite results across levels of behavioral organization are interpreted in the context of intermittent exercise and locomotion whereby decreasing effort at one level of energy expenditure could allow for increased effort at another. Therefore, different (and sometimes opposite) responses to environmental conditions can occur at different levels of behavioral organization. Even when overall foraging effort remains unchanged, the strategy employed can differ in its temporal allocation or frequency, which could be important in balancing energy budgets under increased energetic costs and/or time constraints. These results have particularly important implications for interpreting behavioral responses investigated at only a single level of behavior, extrapolating data from brief observation periods to longer time scales, and foraging models which only consider single levels of behavior such as the dive cycle.