Low occupancy of nest houses by wood ducks (Aix sponsa) does not necessarily indicate an abundance of nest sites in natural cavities, but rather either a low breeding population or nest houses improperly constructed or placed. A low breeding population may be caused by high nest losses to predators, shortcomings in the breeding habitat, or shooting losses. Nest houses which afford superior protection from predators increase numbers of wood ducks, even in habitats with abundant natural cavities. On Quiver Creek in central Illinois, utilization of predator-proof nest houses increased the breeding population of wood ducks from 10-15 pairs to over 90 pairs. Cylindrical metal houses afford greater protection than board houses, but are widely acceptable only when (1) there is a high density of breeding wood ducks, (2) the breeding population is expanding, or (3) a local population is conditioned to nesting in houses. After initial use of metal houses by wood ducks, high nesting success is usually followed by increased usage as homing adults and yearlings swell the local breeding population. Adult wood duck hens usually return to the nesting areas where they last bred successfully. A large proportion of yearling hens return to their former rearing areas. Yearling hens nest later than older hens, and, by observing specific nesting areas and sites used by adults, are influenced to nest nearby and in similar accommodations. The homing of females to a specific nesting area emphasizes the advantage of grouping nest houses in units. Houses should be spaced according to habitat and potential breeding population: two to three per acre in high-quality habitat are recommended. The low rate at which nest houses have been used by wood ducks has disappointed waterfowl biologists in some states. They concluded that enough natural cavities already existed for nesting purposes and that production of young was not increased substantially by providing additional nesting sites. A difference in the rate at which houses were used by wood ducks is also apparent in Illinois, where nesting in houses has been studied continuously since 1938 (Hawkins and Bellrose 1940, Brown and Bellrose 1943, and Bellrose 1955). Through the years, various types of houses have been developed in an effort to reduce nest loss caused by predators (Bellrose 1955). Metal houses grouped in units showed a divergence in use between units that was difficult to evaluate. In 1958, a project was initiated to investigate the relative value of natural cavities and nest houses on a study area in Mason County, Illinois. Nesting of wood ducks in natural cavities was studied in this region from 1938 to 1941, but, for the next 17 years, studies were concerned only with nesting in houses. The principal objectives of this study were to (1) compare the degree of use accorded natural cavities and nesting houses by wood ducks; (2) compare nesting success in natural cavities with that in nesting houses; (3) observe the effect of an increased raccoon (Procyon lotor) population upon nesting success of wood ducks; (4) evaluate the selection of particular natural cavities by nesting wood ducks to improve the dimensions, design, and placement of nest houses; and (5) ascertain whether bands of Tanglefoot or metal strips placed around tree trunks would prevent raccoons from reaching nesting cavities. The original study area extended 15 miles north and 20 miles south of Havana, Mason 1 Present addresses of Kenneth L. Johnson and T. Udell Meyers: Nebraska Game, Fish and Forestation Commission, Lincoln; and U. S. Marine Corps, San Diego, California, respectively.