The American Crow (Corvus brachyrhynchos) because of its gregarious roosting habits lends itself admirably to census methods. In New York State the winter range of the crow falls conveniently into 20 divisions, each of which represents the foraging territory of a population unit. Widely scattered over these foraging territories during the daylight feeding hours, the members of each population unit concentrate their numbers at night at a more or less centrally located roosting site. Here the assembled thousands may be counted and analyzed with comparative ease. The present study, conducted as a graduate research project at Cornell University, brings together information on all the mid-winter roosts located within the state during the 1932-33 season. The information thus obtained has been supplemented with a review of published records in the literature. With this material an attempt has been made to explain the peculiarities of winter crow distribution on an ecological basis. The winter range of the eastern crow (C. b. brachyrhynchos) extends along the Atlantic seaboard from the vicinity of Washington, D. C. north into southern Canada. New York State, the field of the present survey, lies in the northern part yet well within this (winter) range. The largest New York roosts are, with one or two exceptions, as large as any recorded in the eastern states since 1900. The figures and conclusions to be drawn from the present report may therefore be regarded as fairly typical of eastern conditions. Many observers have attempted to count or estimate the numbers of birds congregated in a crow roost. Most of them have failed to produce reliable data for three reasons: 1. The personal factor. The discrepancy between the report of an overenthusiastic amateur and that of an ultra-conservative naturalist often amounts to several thousand per cent. 2. Irregularity of behavior of the birds. A single observation on a roost does not take into consideration the fluctuation in numbers which may take place during the roosting season. For example at the Tompkins County roost in December, 1932, the population in a single week dropped from 15,000 to 3,000 and again rose to 8,000 as a warm spell came and passed. An isolated observation on a roost obviously might be extremely misleading.