Traditionally, transfer is described as interference and consequently as an unintentional mechanism. More recently, however, the perception of control in transfer has changed and it is now commonly accepted that transfer can occur both automatically and strategically. Studies have previously employed think-aloud protocols during writing tasks to establish the degree of intentionality in transfer. However, this method does not let us distinguish between instances of transfer that are truly unintentional and instances that were simply not commented on due to the constraints imposed by the think-aloud protocol. The present study, therefore, conducted a stimulated recall interview in direct succession of the think-aloud protocol in order to categorise also those instances of transfer that were not commented on initially. With data from highly multilingual adult learners of Catalan, the study examines the proportion of unintentional and intentional transfer in relation to (1) different types of transfer, (2) different source languages of transfer and (3) different word classes in transfer. The data indicates that some of these aspects of transfer tend to occur unintentionally, whereas others are more prone to be used intentionally. The theoretical implications of the results, as well as their limitations, are discussed.