Evaluation is viewed as essential to decision making and social policy development. Since conventional methods have been disappointing or inadequate, naturalistic inquiry (N/I) differs from conventional science in minimizing constraints on antecedent conditions (controls) and on output (dependent variables). N/I is phenomenological rather than positivist. It offers alternative strategies for problems when the experimental approach is implausible. A number of new evaluation models (such as the responsive model, the judicial model, and the connoisseurship model) are compatible with the approach. Since there is no compelling way to truth, N/I must be credible and deal convincingly with standard methodological problems such as boundary problems (setting the scope of inquiry), focussing problems (establishing and defining categories), and problems of authenticity (reliability, validity, and objectivity). Techniques for establishing validity include: triangulation, cross examination, persistent observation, and peer or participant corroboration. In evaluation, validity may be ecological, contextual, or phenomenological. Impartiality is imperilled by conscious or unconscious bias, incompetence, gullibility, or corruptibility. It is promoted by openness and fairness. (Author/CTM)