During the 1977 and 1979 reproductive periods of the Galápagos fur seals a census taken in the mornings and evenings at Cabo Hammond, Fernandina, showed a marked, synodic lunar rhythm in numbers of animals ashore. About twice as many fur seals were ashore at full moon than at new moon. By use of two independent Fourier analysis methods, the curve of the morning counts is shown to lag 15°–20° of the lunar month behind the curve of the evening counts. The lunar effect is demonstrated for males, females, and immatures. The rhythm is also seen is demonstrated for males, females, and immatures. The rhythm is also seen in attendance data from 13 individually marked females, all but one nursing young. Reproductive events show the lunar rhythm much less markedly than do numbers ashore. This and the clear rhythm in immature numbers make it very likely that the rhythm is a year-round phenomenon, independent of reproduction. There is no reason to assume that fur seals stay on land during moonlit nights especially for social interaction. It is then hypothesized that fur seals avoid moonlight at sea. If so, the peak of numbers ashore at full moon and the negative phase angle difference of the evening curve against the morning curve can be explained with the shift, and the varying duration and brightness, of the moonlit part of the night over the lunar cycle. Two hypotheses which might account for this moonlight avoidance are discussed: (1) predator (shark) avoidance and (2) varying feeding efficiency of the fur seals due to the influence of moonlight on the vertical distribution of prey.