This article reports meta‐analyses intended to clarify and enhance our understanding of voice and its promotive and prohibitive forms. We find that undifferentiated constructive voice is associated with a wide range of antecedents that fit in Morrison's (2014) five categories: (a) dispositions, (b) job and organizational attitudes and perceptions, (c) emotions, beliefs, and schemas, (d) supervisor and leader behavior, and (e) contextual factors. However, relative weight analyses reveal a highly dominant variable within each category (personal initiative, felt responsibility, engagement, leader–member exchange, and positive workplace climate). We also find that undifferentiated constructive voice has a moderate zero‐order association with job performance that is nonsignificant when task performance and organizational citizenship behavior are also considered. Finally, we explore how associations vary as a function of whether voice is promotive or prohibitive. First, there are significant differences in associations with over a third of the antecedents (core self‐evaluations, felt responsibility, organizational commitment, detachment, psychological safety, ethical leadership, and leader openness). Second, although promotive voice has a positive association with job performance, the opposite is true for prohibitive voice. We conclude with suggestions to enhance our understanding of voice, especially with respect to efforts needed to clarify and distinguish promotive and prohibitive voice.