We examined the contribution of the phonological loop to immediate free recall (IFR) and immediate serial recall (ISR) of lists of between one and 15 words. Following Baddeley (1986, 2000, 2007, 2012), we assumed that visual words could be recoded into the phonological store when presented silently but that recoding would be prevented by concurrent articulation (CA; Experiment 1). We further assumed that the use of the phonological loop would be evidenced by greater serial recall for lists of phonologically dissimilar words relative to lists of phonologically similar words (Experiments 2A and 2B). We found that "in both tasks," (a) CA reduced recall; (b) participants recalled short lists from the start of the list, leading to enhanced forward-ordered recall; (c) participants were increasingly likely to recall longer lists from the end of the list, leading to extended recency effects; (d) there were significant phonological similarity effects in ISR and IFR when both were analyzed using serial recall scoring; (e) these were reduced by free recall scoring and eliminated by CA; and (f) CA but not phonological similarity affected the tendency to initiate recall with the first list item. We conclude that similar mechanisms underpin ISR and IFR. Critically, the phonological loop is not strictly "necessary" for the forward-ordered recall of short lists on both tasks but may "augment" recall by increasing the accessibility of the list items (relative to CA), and in so doing, the order of later items is preserved better in phonologically dissimilar than in phonologically similar lists.