While evaluating the effectiveness of a portion of the public information system at the Toledo Art Museum, nonclassical approaches to data-gathering presented problems in measuring program aspects which are not easily quantified. There is a necessity to clarify what kinds of information can validly be gathered with naturalistic techniques for an audience unfamiliar with evaluation processes but having specific evaluation needs within their organizational activities. Curators and museum educators are seen as looking for evidence of the beneficial effects of the public information program while sound evaluation techniques and professional standards must be maintained. The difficulties of nonclassical evaluation are discussed in terms of problem formulation, the iterative framework for refinement of questions and the participation of the evaluation audience. Precautions for use of naturalistic evaluation are outlined and within the limitations of the techniques in non-traditional settings a payoff is suggested; the kinds of data gathered and reports submitted are often of immediate and practical use to the evaluation audience. Nontraditional forms of evaluation are seen as likely to become increasingly pervasive. (Author/CM)