We studied the breeding of 125 pairs of Common Terns (Sterna hirundo) in a large colony at Monomoy, Massaschusetts, in 1979. The colony was subjected to predation by one or more Great Homed Owls (Bubo virginianus). The adult terns deserted the colony for 6.5-8 hours each night throughout the season. Although the owl(s) took no adults and only about 20 chicks from our study plots, the terns suffered unusually heavy losses from other causes, including breakage and disappearance of eggs, hatching failures, attacks by ants (Lasius neoniger), chilling of newly-hatched chicks, and predation by Black-crowned NightHerons (Nycticorax nycticorax). In a 10-year study, most of these causes of egg and chick loss have been associated with nocturnal desertion and predation by Great Horned Owls. Although nocturnal desertion is effective in minimizing owl predation on adults, it leaves the eggs and chicks vulnerable to chilling and predation. In 1979, both direct and indirect effects of predation fell more heavily on terns that laid in May than on terns that laid in June. Differential predation on early nesters tends to offset other factors that presumably favor early nesting. Predation is one of the most important selective forces influencing breeding behavior in ground-nesting colonial birds, such as gulls and terns (Lack 1968). When a nocturnal predator gains access to a colony, larids usually have no effective defenses (Austin 1948, Ashmole 1963, Southern et al. 1982). However, avian predators and diurnal mammals often take only limited numbers of colonial larids (Hatch 1970, Nisbet 1975, Southern et al. 1982). Consequently, the study of differential mortality in a larid colony can throw light on the role of predation in influencing timing, synchrony, and spacing behavior (Ashmole 1963, Kruuk 1964, Patterson 1965, Tinbergen 1967, Parsons 1971, Nisbet 1975). Nisbet (1975) reported that Great Horned Owls (Bubo virginianus) preying on nesting Common Terns (Sterna hirundo) took a much higher proportion of chicks hatched earlier in the season than of chicks hatched later. We describe here a study of another Common Tern colony subjected to predation by one or more Great Homed Owls. In the case described here, direct predation by the owl(s) was less important than other causes of egg and chick losses. However, we present evidence that at least some of these other losses were indirect consequences of the owl's activity. This paper analyzes seasonal variations in both direct and indirect effects of owl predation. STUDY AREA AND METHODS We studied 125 pairs of Common Terns within a colony of about 3,300 pairs at Monomoy National Wildlife Refuge, Massachusetts (41038'N, 69058'W) in May-August 1979. This was part of a long-term study of this colony, whose results are summarized in Table 1. In 1979, we selected two study plots about 75 m apart, comprising a total area of about 570 m2. Each plot was on the edge of a large open sandy area surrounded by dense beach grass (Ammophila breviligulata). Common Terns nested earliest and most densely in a partially vegetated border strip containing scattered clumps of live and dead beach grass, seaside goldenrod (Soli ago sempervirens) and beach pea (Lathyrus japonica). Although both study plots included areas of dense grass and open sand, Common Terns occupied these habitats later and less densely than the border strips. The plots were selected to be representative of the whole colony, and the density and patterns of occupation appeared similar in other parts of the colony. We visited the study plots on most days from 22 May to 9 July, on 20, 21 and 28 July, and on 7 August 1979. P. Trull visited the colony n five other days between 10 and 25 July. N ts and eggs were marked when first seen, and eggs were weighed. When eggs were not found on the day of laying, this date was estimated from laying and hatching patterns and by flotation (Hays and LeCroy 1971); weights of fresh eggs were estimated from data on the rate of weight loss (Rahn et al. 1976). Nine banded adults, including two nesting outside the plots, were trapped to determine their ages. Chicks were banded at hatching and checked