Crossmodal correspondences have been widely demonstrated, although mechanisms that stand behind the phenomenon have not been fully established yet. According to the Evaluative similarity hypothesis crossmodal correspondences are influenced by evaluative (affective) similarity of stimuli from different sensory modalities (Jankovic, 2010, Journal of Vision, 10(7), 859). From this view, detection of similar evaluative information in stimulation from different sensory modalities facilitates crossmodal correspondences and multisensory integration. The aim of this study was to explore the evaluative similarity hypothesis of crossmodal correspondences in children. In Experiment 1 two groups of participants (nine-and thirteen-year-olds) were asked to make explicit matches between presented auditory stimuli (1 second long sound clips) and abstract visual patterns. In Experiment 2 the same participants judged abstract visual patterns and auditory stimuli on the set of evaluative attributes measuring affective valence and arousal. The results showed that crossmodal correspondences are mostly influenced by evaluative similarity of visual and auditory stimuli in both age groups. The most frequently matched were visual and auditory stimuli congruent in both valence and arousal, followed by stimuli congruent in valence, and finally stimuli congruent in arousal. Evaluatively incongruent stimuli demonstrated low crossmodal associations especially in older group.