The idea that assessment is intrinsic to effective instruction is traced from early experiments in the individualization of learning through the work of Benjamin Bloom to reviews of the impact of feedback on learners in classrooms. While many of these reviews detailed the adverse impact of assessment on learning, they also indicated that under certain conditions assessment had considerable potential to enhance learning. It is shown that understanding the impact that assessment has on learning requires a broader focus than the feedback intervention itself, particularly the learner's responses to the feedback, and the learning milieu in which the feedback operates. Different definitions of the terms "formative assessment" and "assessment for learning" are discussed, and subsumed within a broad definition that focuses on the extent to which instructional decisions are supported by evidence. The paper concludes by exploring some of the consequences of this definition for classroom practice. (Contains 1 figure and 3 tables.)