A study was conducted of the capacity of sixth-grade children to communicate accurately paralinguistic affect. A dependent measure indicating the accuracy of paralinguistic communication of affect was obtained by comparing the level of affect which children intended to encode with ratings of vocal inflections from adult judges. Four independent variables were manipulated: (1) the sex of the child doing the encoding (male/female), (2) the literal content of the child's verbal message (positive/negative), (3) the encoder's intended level of vocal intonation (extremely unfriendly, moderately unfriendly, neutral, moderately friendly, and extremely friendly), and (4) the sex of the adult decoder who provided ratings of vocal affect conveyed (male/female). Four hypotheses were tested. The hypothesis that paralinguistic encoding should be more accurate when literal content and intended affect are consistent, rather than inconsistent, was partially confirmed. The hypothesis that negative paralinguistic inconsistency should be more accurately encoded than positive paralinguistic inconsistency was also partially confirmed. Contrary to prediction, male children were more accurate encoders than female children. Finally, as predicted, adult females were more accurate decoders than adult males. Findings are discussed in light of the role of physiological arousal in affecting motor performance according to learning theory. (Author)