In much previous research, listeners' rating data have served as a dependent variable to demonstrate the effects of age of learning, length of residence, and motivation on L2 users' degree of foreign accent. However, the role of speaking rate in such judgments has not been ascertained. To gain new insight into this relationship, we carried out two experiments involving sentence-length utterances produced by English L2 users. In the first, we observed a significant curvilinear relationship between speaking rates and accentedness and comprehensibility judgments of utterances produced by users from a variety of L1 backgrounds. In the second experiment, by manipulating rates with speech compression-expansion software, we established that this effect was due to the rate differences themselves, rather than to differences in L2 proficiency that might co-vary with rate. In both experiments the listeners tended to assign the highest ratings to L2 speech that was somewhat faster than the rates generally used by L2 users; however, both very fast and very slow speech tended to be less highly rated. Researchers who use listener rating data should be mindful of the potential confounding effect of speaking rate in their data.