The aim of the reported experiment was to obtain insight into how learners' visuo-spatial working memory is involved during learning how to perform procedural-motor tasks from a multimedia instruction (i.e. 'learning how'). Eighty-two participants studied first-aid procedures using text only or multimedia. Working memory involvement was gauged by measuring the interference between learning first-aid procedures and performing a spatial dual task. Learning outcomes were measured as task performance and task description. The results showed that performing a spatial dual task interfered to a larger extent with learning from text only than from multimedia. The results tend to support the assumption that pictures in tasks focusing on 'learning how' are beneficial to learning, because they might omit the need to engage in imagery and therewith reduce the cognitive effort that is required to understand the learning material. © 2013 John Wiley & Sons, Ltd.