Three experiments are reported which investigate the application of the directed forgetting task to emotionally valent material and to different mood states. In this task, some subjects are told when halfway through an intentional or incidental learning task that the trials presented so far are to be forgotten because they were merely practice. However, at the end of the subsequent list, the subjects are then asked to recall all of the previous items including those that they were instructed to forget. The results typically show that significantly fewer directed forgetting items are recalled whether the task is an intentional or incidental learning one. In the first experiment, normal and 'depressed' students rated positive and negative material for pleasantness; although directed forgetting effects were obtained, there were no differential effects of mood state nor of valence of the material. In order to investigate this effect further, a variant of this task was used in Experiment 2 in which the positive and negative material had to be processed in relation to the self. The results showed that differential forgetting now occurred; healthy students recalled more positive than negative information, but this positive bias was not obtained for 'depressed' students who showed an even-handed level of recall. In Experiment 3, groups of clinically depressed, clinically anxious, and normal controls were presented with the directed forgetting task. The key finding showed that the depressed subjects showed a retrieval facilitation for to-be-forgotten negative adjectives, an effect that was not present for the other two groups. It is concluded therefore, that the directed forgetting task could be usefully extended to investigate cognition-emotion interactions in clinical populations.