The current experiments examined how native Parisian French and native Swiss French listeners use vowel duration in perceiving the /[openo]/-/o/ contrast. In both Parisian and Swiss French /o/ is longer than /[openo]/, but the difference is relatively large in Swiss French and quite small in Parisian French. In Experiment 1 we found a parallel effect in perception. For native listeners of both dialects, the perceived best exemplars of /o/ were longer than those of /[openo]/. However, there was a substantial difference in best-exemplar duration for /[openo]/ and /o/ for Swiss French listeners, but only a small difference in best-exemplar duration for Parisian French listeners. In Experiment 2 we found that this precise pattern depended not only on the native dialect of the listeners, but also on whether the stimuli being judged had the detailed acoustic characteristics of the native dialect. These findings indicate that listeners use fine-grained information in the speech signal in a dialect-specific manner when mapping the acoustic signal onto vowel categories of their language. (Contains 2 figures, 2 tables and 4 notes.)