This paper presents a production study of incomplete neutralization in American English flapping. In flapping, /d/ and /t/ both become a voiced flap in certain prosodic contexts (see, e.g., Kahn 1980). A number of studies show that this neutralization is incomplete: /d/-flaps can be distinguished from /t/-flaps on the surface (Fox and Terbeek 1977). Other studies, however, have found conflicting results (Port 1976). This study finds that flapping is an incompletely neutralizing process—/d/-flaps and /t/-flaps can be distinguished on the surface by the duration of the preceding vowel, at least for some speakers. Additionally, some studies find evidence that hyperarticulation and orthography have an effect on whether neutralization is complete or incomplete (Fourakis and Iverson 1984, Warner et al. 2006). The present study employed two tasks: a minimal pair reading task, designed to increase these potential effects, and a morphological paradigm completion/"wug" task, designed to reduce these effects. No significant differences between the two tasks were found, thus failing to support the claim that incomplete neutralization is due to these extragrammatical factors.