The 1992 Teacher Follow-up Survey (TFS) Reinterview and Extensive Reconciliation represents the Bureau's first attempt to employ an extensive structured reconciliation. Cognitive methodology was utilized to develop the reconciliation questions, which were meant to elicit why respondents' answers differed between the original and the reinterview. The ultimate goal was to identify instances in which respondents' answers differed because they were having difficulty comprehending the questions or the response tasks asked of them. Respondents often reported not knowing why their answers differed, or that they misunderstood the questions or thought the answer categories were confusing. This paper offers suggestions for improving a number of the TFS questions based on these results. It also makes suggestions for improving the reinterview and reconciliation methodology.