Three experiments examine why the recall of performed actions often fails to profit from provision of retrieval context. A cue failure has, for instance, been demonstrated with cued recall for performed actions in that cued recall is lower than free recall, whereas control conditions show the usual free to cued recall increase for non-enacted material. The first experiment confirms the cue-failure effect and extends the generality of the finding to everyday cue contexts (where the cues are represented by images of the locations associated to the actions), which intuitively should be of general retrieval help. The second experiment shows that the cue-failure effect is also present, even to a greater extent, in congenitally totally blind people; the third experiment suggests how the cue-failure effect may be defeated by means of simultaneous motor and non-motor encoding. The results are discussed in terms of strategic and non-strategic conceptions of action memory and of the independence of motor and visuospatial codes. This independence appears maintained also in the blind.