A migration of the diurnal moth Urania fulgens showed unusual timing. Unlike almost all other recorded migrations, this one took place in the dry season and consisted entirely of females with large batches of mature eggs. Our data suggest that the migrant/non-migrant ratio may differ among migrations, and migrations may take place for different reasons, some of which may relate to seasonally modulated availability of larval food plants. CLIMATIC STABILITY IN THE TROPICS is assumed by several theories which seek to explain the extraordinary diversity of species in tropical areas (e.g., Slobodkin and Sanders 1969, MacArthur 1969), yet there is a persistent controversy over whether tropical areas are more stable and predictable than those of higher latitudes (Cain 1969, Preston 1969, Smith 1972). Many insect species show violent population fluctuations and vast movements in the tropics (Williams 1958), and understanding them should shed some light on the degree of stability of tropical ecosystems. The diurnal moth genus Urania has been known for nearly two hundred years to be migratory (Tutt 1902; Williams 1930, 1958; Valerio 1966; Chemsak et al. 1979), yet very little is known about the details of its migrations (Williams 1958) or its general ecology (Valerio 1966, Young 1970, Smith 1982). In this paper we describe a migration of Urania fulgens that differs in several important aspects from previously reported migrations. Urania fulgens was observed in three situations in a lowland tropical forest of Costa Rica: as solitary individuals, in aggregations, and as migrants. For four weeks prior and three weeks after the migration described here, only ten solitary moths and three aggregations of less than 100 moths each were found in a radius of about five kilometers around La Sirena in the Parque Nacional Corcovado. From 23 to 26 January 1983, an entire migration was observed at La Sirena in the Parque Nacional Corcovado. The observation of a complete migration is a rare opportunity due to their usual great geographic extent (Fig. 1) and long duration (e.g., Williams 1958, Young 1970, Smith 1972, Chemsak et al. 1979). MATERIALS AND METHODS Ten solitary moths were collected between 23 December and 25 January. On 1 January, 13 individuals were collected from a roosting aggregation, and 50 moths were collected from the migration. Moths were sexed, and when possible their wings inspected for wear. Following Watt et al. (1977) and Tabashnik (1980), freshly emerged specimens were given a wing-wear rating of 0, while very tattered individuals were given a value of 5. To minimize individual bias in evaluating condition, only one researcher rated wings. Since specimens had to be rated at capture, this resulted in some missing values. Eggs in females were counted (or estimated where numbers were high) and the bursa copulatrix examined for spermatophores to ascertain whether the individual had mated. To estimate the size of the migrating population, a 50 m transect was defined across the flight path and moths flying through it were counted over set periods of time.