Competency-based education (CBE) programs are growing in popularity as an alternative path to a postsecondary degree. Freed from the seat-time constraints of traditional higher education programs, CBE students can progress at their own pace and complete their postsecondary education having gained relevant and demonstrable skills. The CBE model has proven particularly attractive for nontraditional students juggling work and family commitments that make conventional higher education class schedules unrealistic. But the long-term viability of CBE programs hinges on the credibility of these programs' credentials in the eyes of employers. That credibility, in turn, depends on the quality of the assessments CBE programs use to decide who earns a credential. In this paper the authors introduce a set of best practices for high-stakes assessment in CBE, drawing from both the educational-measurement literature and current practices in prior-learning and CBE assessment. Broadly speaking, there are two areas in assessment design and implementation that require significant and sustained attention from test developers and program administrators: (1) validating the assessment instrument itself and (2) setting meaningful competency thresholds based on multiple sources of evidence. Both areas are critical for supporting the legitimacy and value of CBE credentials in the marketplace. This paper therefore details how providers can work to validate their assessments and establish performance levels that map to real-world mastery, paying particular attention to the kinds of research and development common in other areas of assessment. The authors also provide illustrative examples of these concepts from prior-learning assessments (for example, Advanced Placement exams) and existing CBE programs.