Sensitivity to feature co-occurrences was investigated as a function of analytic and analogical transfer. Participants memorized descriptions of hypothetical people and were then induced either to make transfer decisions by analogy to the descriptions (analogical transfer) or to search for and apply rules (analytic transfer). Across three experiments, analogical transfer was found to be more effective than analytic transfer for preserving co-occurring features in classification judgements. This result held for a variety of category structures and stimulus materials. It was difficult for subjects who adopted an analytic-transfer strategy to identify regularities that were embedded in stored instances. Alternatively, subjects who adopted an analogical-transfer strategy preserved feature co-occurrences as an indirect result of similarity-based retrieval and comparison processes. The effectiveness of analytic and analogical transfer is discussed.