Acoustic characteristics of affective prosody in 30 children aged between 36 and 48 months, matched for SES and IQ, were measured in spontaneous and imitated utterances. Intonation patterns for neutral-normal, happy, sad, and angry affect were elicited and audiorecorded under controlled conditions. The responses were transcribed phonetically to account for speech variations relevant in the analysis. The recordings were analyzed acoustically for characteristics of fundamental frequency, intensity, duration, and their co-variation for each affective pattern using Oscillomink (Siemens) tracings and a PM pitch analyzer (voice identification). It was hypothesized that the production of affective prosody would parallel its perception. An earlier study (C. Baltaxe, Proceedings of the Sixth International Phonology Meeting, 1–7 July 1988, Krems, Austria) had shown that the same age group perceived the above affective patterns beyond chance. However, degree of success was linked to specific emotion, with a Polyanna effect toward happy emotion. The results of the present study showed that affective productions did not parallel perception, even in the imitation condition. Results are discussed with respect to articulatory control necessary for each pattern. [Work supported by Maternal Child Health.]