The present experiment examined the decoding and encoding of conflict messages in order to assess the relative contribution of facial expression, tone of voice, voice, and content of speech to the integrated impression of the whole. Undergraduate and graduate students (N=16, aged 20-28) were required to study acted, videotaped conflicting messages. The results were as follows: 1) Facial expression of encoders played a major role in decoding of conflicting messages. However, in decoding negative facial expression, the voice in contributing more than any others to interpret the integrated impressions of the whole. 2) Males were more consistent than females in decoding affect of the encoders. 3) Encoders themselves rated the contribution of vocal and visual channels equally. However, negative facial expression was evaluated disproportionally negative to the original intention of encoders by decoders. These results suggested that the Japanese university students tend to shift their point of evaluation to another point when encountering negative facial expressions in order to avoid negative social interactions while decoding.