Recent years have seen a proliferation of evidence that simple pairings of brands with affective stimuli (e.g., emotional appeals in advertisements) can have uncontrolled effects on people’s brand attitudes, known as uncontrolled evaluative conditioning (EC) effects. This raises important ethical questions regarding consumers’ autonomy and the regulation of commonly used advertising techniques. In the current research, we investigate for the first time whether visual affective stimuli (pictures) differ from verbal affective stimuli (words) in the extent to which they generate controlled and uncontrolled valence transfer. As the processing differences between verbal and visual stimuli are multidimensional in nature, it is challenging to determine which type of stimuli will have more controlled or uncontrolled effects on attitudes. Three experiments consistently (and from some theoretical perspectives, unexpectedly) demonstrate that valence transfer from visual stimuli is easier to control than valence transfer from verbal stimuli. Furthermore, once control fails, verbal and visual affective stimuli appear equally effective at uncontrolled valence transfer. Theoretical as well as practical implications of our findings are discussed.