Many studies over the past 50 years have sought to identify environmental factors influencing breeding duck abundance and distribution in northern North America. Because results are currently scattered within the scientific literature, a summary of established duck-habitat associations would help to orientate future modelling research. Our goal was to review the published research testing for duck-habitat associations in northern North America. We reviewed 124 studies, summarizing their geographical coverage and species representation, and then analyzing the duck-habitat associations they tested. We identified 533 associations on 133 covariates falling into 16 environmental classes. Covariates of the ‘wetland' classes were the most frequently associated with ducks; among these, ‘wetland area' and ‘wetland density' were the most common. ‘Climate’ covariates were the second most common associations, suggesting the potential for projecting the effects of climate change on ducks. The best-documented anthropogenic class was ‘agriculture’, for which associations with ducks were mostly negative. However, relatively few studies tested for associations with covariates for anthropogenic disturbances, which suggests that more research is needed to support forecasts of duck distribution under future human activity. This review and the accompanying database of duck-habitat associations will support future modelling studies by facilitating the selection of suitable habitat covariates.