Abstract Traditionally, recovered memories of childhood sexual abuse (CSA) have been classified as those emerging spontaneously versus those surfacing during the course of suggestive therapy. There are indications that reinterpretation of memories might be a third route to recovered memories. Thus, recovered memories do not form a homogeneous category. Nevertheless, the conceptual distinctions between the various types of recovered memories remain difficult for researchers and clinicians. With this in mind, the current study explored whether recovered memories can be reliably classified. We found that classification is rather problematic in a subset of cases. To reduce potential bias, we urge for the development and subsequent use of a more reliable classification system and multiple expert raters in research on recovered memories.