In a modified replication study, a general question intended to capture spontaneous detection of inconsistent information found neither good nor poor comprehenders successful. In response to a specific question, however, good comprehenders evinced far superior error detection than poor comprehenders. Reasons are speculated upon. (Author/MT)