Seven experiments are described investigating the effect of high intensity white noise during the visual presentation of words on a number of short-term memory tasks. The findings were: 1. In a free recall task recall of items decreased at the highest intensity used (85 dB) compared with a quiet and a 75 dB condition. 2. In free recall, recall by category decreased and recall in the original sequence increased in the 75 dB compared with the other two conditions. 3. Recall of the position of words in the list increased as noise intensity increased, but only when the learning of position was incidental, not when it was intentional. It is inferred that the effect is due to direction of attention or change in the learning strategy. 4. Recall of the original sequence (as shown by the ability to give in response to a word from a list the word which had followed it in the original list) was superior in the 75 dB compared with the other two conditions, but only when recall of the second word was required, not when it had to be recognized among all the items from the original list. It is argued that this can be explained if noise intensity affects the strength of traces and hence the interconnexions established between them, on which retrieval depends. The results for position learning are compatible with the theories of Hockey & Hamilton (1970) or Dornic (1973), but the results for sequence learning cannot be explained by either of these theories. A final experiment confirmed a prediction from the above theory that when recalling the original sequence, omissions (recalling no word) will decrease and transpositions (giving the wrong word) will increase as noise level increases.