After a discussion of labelling and meaning, a diagrammatic model was proposed to illustrate possible encoding of verbal and pictorial information. Two experiments using visual stimuli were carried out to investigate the relative likelihood of these encodings. An attempt was made to equate the number of attributes encoded from each type of stimuli. Questionnaires were used to provide information on encoding. Experiment 1 required retention of a short list; verbal and pictorial stimuli were remembered equally well. Verbal encoding was predominant for both types of stimuli and was related to better recognition whereas visual imagery appeared to be non-functional. Experiment 2 required recall of one stimulus after an interpolated task; recall was much superior for pictorial stimuli. This was thought to be because pictorial stimuli were encoded twice, verbally and visually, to a greater extent than were verbal stimuli. Both forms of encoding were positively related to recall. These results, in conjunction with those from other studies, suggest that verbal encoding is particularly useful for sequential information.