Adger (2006) argues that morphosyntactic variability within the speech of an individual can be captured in an approach where competing variants have uninterpretable features that may be checked in the same syntactic context. The results of a written survey of 90 native speakers of English suggest that the distribution of pronoun case forms in coordinates (1)-(3) and other strong pronoun contexts (4)-(5) exhibits exactly the kind of non-deterministic variability Adger’s approach was designed to capture. (1) [He/him and I/me] arrived here three hours ago. (2) Brenda had promised she would meet [he/him and I/me] at the station. (3) The landscapes painted by [he/him and I/me] drew huge crowds at the exhibition. (4) [We/us New Zealanders] must stick together. (5) It was [he/him] who insisted on going to the rally. In this paper, I outline a feature-based analysis that accounts for the most commonly attested pronoun case patterns and provides supporting evidence for Sigurðsson’s (in press) claim that ‘uninterpretable’ features actually serve to interpret abstract syntactic and semantic relations at the PF(=Phonetic Form)-interface.