Two large-scale (25 ha) trials were initiated in 1989 in Wisconsin to compare six alternative production systems regarding productivity, profitability, and environmental impact. The project was designed and is managed by a coalition of farmers, extension agents and research personnel. Deliberations between production-oriented and ecologically oriented team members resulted in a factorial design, with two enterprise types (cash grain and forage-livestock) and three levels of biological complexity. Statistical methods have been used to identify the most efficient plot size, plot shape, and block shape, and the optimal procedures for sampling soil characteristics. A uniformity year was allowed before initiation of the trial, and the start was staggered. We defined treatments as production strategies rather than a specific set of inputs, which led to a more flexible plot management program.