The retention of trees in harvest units is an integral part of forest management practices on federal lands in the northwestern United States (U.S.), yet the ecological benefits that result from various levels or patterns of retained trees remain speculative. Large-scale and long-term silvicultural experiments are needed to evaluate the effects of alternative forest management strategies on biological diversity, yet they are rarely undertaken due to the substantial commitments of time and resources required. The Demonstration of Ecosystem Management Options (DEMO) study was initiated by U.S. Congressional mandate in 1993 and is unique in the northwestern U.S. in its geographic scope, experimental design, and collaboration between research and management. The experimental design consists of six harvest treatments, each 13 ha in size, replicated at six locations (blocks). Treatments represent strong contrasts in the level (15-100%) and pattern (dispersed or aggregated) of retention in mature Douglas-fir forests. Data were collected from 1994 to 1996 prior to harvest, experimental harvests were implemented in 1997 and 1998, and initial post-treatment sampling was completed in 2001. A variety of ecological responses were measured and public perceptions of visual quality were evaluated. We discuss the objectives and experimental and sampling designs for the DEMO study, and present brief summaries of the results of component studies. Early results have provided important insights into response mechanisms, but the most significant and useful results will be obtained as experimental stands reach maturity in the coming decades.