Canada geese (Branta canadensis) are unusual because individuals use either of two different brood-rearing behaviors: cooperative broods (two or more merged broods attended by more than three parents) or two-parent families. We tested whether cooperative broods form in response to habitat or climatic conditions by examining variation in cooperative brood frequencies among Canada geese nesting in Connecticut from 1982 to 1996. Percent of goslings raised in cooperative broods ranged from 0 to 100% at a given site in different years, but the pattern of annual variation was different at each site. The sites were in close proximity to each other and had similar climates; thus, the differences in annual variation among sites was not likely to be a response to climatic conditions. Cooperative brood frequencies also varied among sites in each individual year, but sites with the most gang brooding in one year often had the least the next. Such would not be expected if gang brooding occurred in response to non-ephemeral habitat characteristics. Sites where gang brooding occurred and where it did not have similar food resources and predation risks. These findings failed to support the hypotheses that gang broods form in response to food competition or predation. Sites where gang broods occurred had more parent geese and more goslings than sites where they did not occur. Furthermore, the proportion of goslings raised in gang broods was correlated with the number of goslings and parents at the site. Our results support the hypothesis that gang broods form from the inadvertent mixing of goslings. This single factor, however, was not sufficient to account for all of the observed variation in gang brooding frequencies.