Paap and Noel's (1991) recent demonstration that increased memory load facilitates naming of low-frequency irregular words has been taken as strong support for dual-route theories of word naming (e.g., Coltheart, 1978). Others, however, (Jared, personal communication, March 1993; Strain, personal communication, April 1993) have been unable to replicate this effect. In the present research, five different attempts were made to determine the crucial experimental conditions required for successful replication of the Paap and Noel findings. None of these experiments produced results at all similar to those reported by Paap and Noel. A number of explanations for these failures to replicate, in terms of individual differences between Paap and Noel's subjects and the present subjects, were evaluated. Most of these explanations were not supported by the data. The present results call into question the generalizability of the Paap and Noel findings and, hence, the strength of support they provide for dual-route theories.