The methods used to determine family composition and colony size in Eurasian beaverCastor fiber Linnaeus, 1758 are often poorly described in published reports. Here we show how repeated counts of colony size in a random sample of colonies (n = 19) varied between dusk and dawn, between the months of August and September, and following successive counts. Mean counts at dusk and the following dawn did not vary significantly, though mean colony size was significantly greater in August than September. However, because all colony members are rarely seen during a single dusk or dawn count, successive counts often provided new information about the maximum number in each age class. This allowed us to adjust colony size following each count using the largest values thus far obtained, with the result that the mean adjusted colony size increased during six successive counts over seven weeks from 2.4 to 3.8. Family composition based on information from all six counts was 54% adults, 26% yearlings and 19% kits. Evidence suggests that kits in particular are undercounted by this method. Figures for colony size and composition in beaver should be viewed with caution if not obtained by methods tested for both precision and accuracy.