Two experiments are reported here in which Ss were asked to rate videotapes of a performer reading friendly, neutral and hostile messages in a friendly, neutral or hostile non-verbal style. These messages and non-verbal styles had previously been presented independently to a separate group of Ss jor rating, in order to obtain an estimate of their individual strengths in terms of six rating scales, and thus permit a matching of verbal (messages) and non-verbal (styles) cues in the experiment where both types of cues were presented in combination. The results of both experiments indicate that non-verbal cues had a greater effect on ratings made on 7-point scales, such as hostile-friendly, than verbal cues. The magnitude of this greater effect of non-verbal cues, however, was dependent on the relative strength of non-verbal as opposed to verbal cues. In the first experiment, both types of cues were approximately equal in strength when ruted alone; here non-verbal cues accounted for 12.5 times us much variance us verbal cues, and produced 5.7 times as much shift on the ratitig scales. In the second experiment the verbal cues were much stronger than the non-verbal cues when rated alone. Here the relative effect of non-verbal cues in the second experiment was diminished; the ratio of non-verbal : verbal variance was now 1.67:1. When verbal and non-verbal signals were inconsistent, the performance was rated as insincere, unstable and confusing - which was not found in earlier experiments on the superior-inferior dimension.