After reviewing the classical approaches of input, process, and outcome evaluation, the discussion focuses on three complementary quantitatively oriented methods for evaluation that can be used in conjunction with one or more of the classical approaches. The complementary methods are categorized as Bayesian approaches, adaptive methods of evaluation, and model-based methods. The analytical techniques described are advised to work best in precisely structured environments with readily measurable input, process, and outcome variables. They tend to become analytically intractable when imbedded in an imprecisely formulated environment. While this would appear to limit the applicability of modern quantitative techniques within the imprecise field of evaluation, the careful and judicious use of abstractions in evaluation settings can yield valuable insights for the evaluation designer and implementer to improve understanding of the many facets of evaluation. The alternative is to ignore the need for comprehensive conceptualizations of the evaluation enterprise. Decision-oriented approaches to evaluation show considerable promise for reducing the misallocation of expensive evaluation resources, the collection of redundant information, haphazard responses to unexpected changes in the program, rote performance of statistical analyses without sequentially formulating and testing hypotheses, analysis and display of information without regard to the decisions to be influenced by it, and the tendency to view rigid experimental design as the ultimate paradigm for evaluation. Fifteen references are provided.