Despite the best of intentions, qualitative researchers can be faced, in some circumstances, with having to make meaning from thin, or less than optimal, data. Using a real study as context, the authors describe the ways that they made sense of their thin data on teachers' perceptions of a large-scale evaluation instrument. They propose a rigorous, sequential approach to e-mail interview analysis based on three hierarchical language levels--lexical, semantic, and pragmatic--to save, by analysis, meanings that might otherwise have been lost because of limited data availability. Researchers are, of course, cautioned to avoid making claims that cannot be substantiated by the data. (Contains 4 tables and 2 figures.)