This paper presents a critical analysis of two recent studies on error correction, Ellis, Sheen, Murakami, and Takashima (2008) and Bitchener (2008) and argues that neither study constitutes evidence for error correction based on discussion of the comparability of participants, design issues, and analysis flaws in both studies. The paper further suggests that research into the efficacy of error correction should focus on linguistic features that are more treatable but less teachable if this practice is to hold out any pedagogical promise. (Contains 2 tables.)