The evidence for qualitatively distinct domains of processing was examined using both recall and recognition measures. The study involved a test of the prediction made by Craik and Tulving (1975) that any semantic analysis would be more beneficial than even a complex physical analysis. In addition, a test was made of processing load involved in tasks representative of the processing domains. Although the strong version of the domain hypothesis proposed by Craik and Tulving (1975) was supported for recognition, there were no differences in levels of recall or recognition for physical and phonemic tasks. Recognition for words on an unattended list was at similar levels for all tasks, suggesting that the tasks did not differ with respect to processing load. Problems with the domains of processing hypothesis are discussed.