Memory for words rated according to their relevance in a grassland survival context is exceptionally good. According to Nairne, Thompson, and Pandeirada's (2007) evolutionary-based explanation, natural selection processes have tuned the human memory system to prioritize the processing of fitness-relevant information. The survival-processing memory advantage has been replicated numerous times, but very little is known about the proximate mechanisms behind it. The richness-of-encoding hypothesis (Kroneisen & Erdfelder, 2011) implies that rating the usefulness of items in a survival context leads to the generation of a large number of ideas that may be used as retrieval cues at test to boost recall. In Experiment 1, the typical survival-processing recall advantage was obtained when words were rated according to their usefulness in 1 of 3 fictional contexts. In Experiment 2, participants were asked to write down any ideas that came to mind when thinking about the usefulness of the words. Consistent with the richness-of-encoding hypothesis, participants generated more ideas in the survival condition than in the fitness-irrelevant control conditions. In Experiment 3, participants generated more ideas for congruent than for incongruent words, demonstrating that the richness-of-encoding hypothesis can also account for the previously obtained congruency effect on recall (Butler, Kang, & Roediger, 2009). In both experiments the number of ideas written down predicted future recall performance well. Our results provide further evidence for the assumption that richness of encoding is an important proximate mechanism involved in memory performance in the survival-processing paradigm.