This study examined the features of cross-modal Stroop interference in picture-word processing by making comparison between auditory-visual and visual-visual tasks. Eleven subjects participated in each task and were instructed to name the picture as quickly as possible. To clarify the semantic-relatedness effect on the amount of interference, five types of different degrees of picture-word semantic relation were used: same stimulus (SS), same category (SC; both items belong to the same category), incongruent (IC), neutral (N) and control condition (C). To investigate the time course of information processing, stimulus onset asynchrony between the word and the picture were varied. In auditory-visual task strong interference effects were obtained in SC, IC and N conditions when the words preceded the pictures by 200,100 and 0 ms. In visual-visual task the similar effects were shown in SC and IC conditions within a 100 ms envelope around simultaneous onset of the two components. These results were discussed in relation to the locus of Stroop interference and the processing mechanisms of picture naming and word reading.