Over the past four decades, there has been much debate on key sources of data in evaluating education, determining school effectiveness, and providing evidence to inform accountability and education planning. Entangled in this debate has been the extent to which large-scale assessments of learning provide valid evidence about the quality of schooling and education in Canada and how they can be used to inform education practice and policy. This article discusses five issues in large-scale assessments that are key for their usefulness and for making valid inferences. Based on recent research on assessment design and validity, the authors offer recommendations for large-scale assessments to better serve the multiple purposes they are intended to serve.