Self-recordings, when speakers record themselves without a researcher present, are attractive for potentially eliciting a wider range of styles than is obtained through interviews. To compare the stylistic differences between self-recorded speech and interview speech, we present an analysis of sibilant production among four speakers in both contexts. Our results show that the contrast between self-recordings and interviews can be a reliable predictor, with differences often surpassing those between interview speech and read speech. We suggest that self-recordings may be stylistically different enough from interviews to justify overcoming the practical challenges of their collection, integrating the self-recording into standard sociolinguistic methodologies, at least for studies of intraspeaker variation and the description of variable phenomena.