Recent research has shown that the ability to inhibit irrelevant information is related to Working Memory (WM) capacity. In three experiments, we explored this relationship by using a list-method directed-forgetting task (DF) in participants varying in WM capacity. Contrary to predictions, in Experiment 1, DF effects were only found for participants with low WM capacity, whereas high WM capacity participants did not show this effect. This unexpected pattern of results was explained as due to the differential susceptibility to interference of high and low span participants. In Experiments 2 and 3, interference was increased by introducing a memory load between the to-be-forgotten and the to-be-remembered list (Experiment 2) and by increasing the list length (Experiment 3). In these conditions of high interference, the pattern of results was reversed so that DF effects were obtained for the high span group and they were not present for the low span group. The reversal of the effect for the high and low WM capacity group depending on the degree of interference suggests that inhibition in the DF procedure depends on both: the degree of interference experienced by the participants, and the availability of controlled resources.