This study examined longitudinal and cross-sectional age effects on accuracy of decoding nonverbal cues. A videotaped nonverbal discrepancy test was administered to children aged 9 to 15 years. The discrepancy test measured: (a) decoding accuracy-the extent to which subjects were able to identify affects (positivity and dominance) from video (facial and body) cues and audio (content filtered and random spliced) cues; (b) discrepancy accuracy-the extent to which subjects recognized the degree of discrepancy between audio and video cues; and (c) video primacy-the extent to which subjects were more influenced by video (face or body) than by audio cues. It was found that (a) older children benefited more than younger ones from the effects of retesting in their accuracy at decoding discrepant cues, especially for discrepant facial cues; (b) all children showed significantly less video primacy after retesting, and older relative to younger children displayed a trend for less body primacy after retesting; and finally (c) relative to younger children, older children showed less video primacy in decoding extremely discrepant ("leaky"), as compared to slightly discrepant, audio and video nonverbal cues. Thus, although older children performed better than younger children at most nonverbal decoding tasks, the advantages of age were especially great for the decoding of the more discrepant or leakier channels. The processes that may underlie increases in specific nonverbal skills and changes in nonverbal styles are discussed.