Two experiments tested the hypothesis that color-naming speed is slower for inconsistent than for consistent color-word combinations as defined by separate semantic differential ratings of color and word components. In Exp. 1 15 undergraduate subjects individually made telegraph-key identifications of consistent and inconsistent instances of yellow or brown color-word combinations. For Exp. 2 and analogue of the Stroop Color-Word Test was constructed in which two cards contained only consistent, and two cards contained only inconsistent, adjective-color combinations. Autonym adjectival pairs, e.g., "hot"--"cold," were used ensuring that two card-pairs contained cards identical in every respect except consistency. Color-naming times of 40 undergraduate subjects for consistent and inconsistent cards were compared. Results of the two experiments supported the hypothesis. Since semantic differential ratings presumably index implicit reactions to stimuli having no obvious relationship to color-naming, these data suggested that a response-competion explanation of semantic-color interaction is incomplete.