This paper presents a rationale for taking a developmental perspective towards evaluation of early childhood programs and contrasts two models of development (organismic and mechanistic) with regard to evaluation and developmental issues. Examination of theoretical differences between the organismic model (which assumes an active organism) and mechanistic model (which assumes a reactive organism) suggests that a developmental perspective: (1) implies that behavioral observations must be analyzed with respect to their meaning in context; (2) raises the question of the meaningfulness of educational testing situations; (3) is important in the conceptualization of development in early childhood programs; and (4) leads to a shift away from a preoccupation with criterion performance to a concern with the processes underlying performance. A developmental perspective also emphasizes: changes in cognitive structures rather than behavioral change per se; discontinuity rather than continuity of change; reciprocal causality between organism and environment rather than unidirectional causality of environment on organism; and organized complexity rather than linear causality in change. The implications of these model features for assessing program effectiveness are discussed. The rationale presented for working toward a developmental perspective in evaluation highlights the involvement of participants in understanding program development, adult (staff and parent) development, and child development. (JMB)