Research with the remember-know paradigm into the subjective experience that accompanies retrieving information from memory has been conducted with the aim of explaining the phenomenon of retrieval experience per se, or with the aim of explaining recognition memory performance, assuming that retrieval experience allows for separating different memory systems. In both lines of research, conclusions are typically based on the rates of remember and know responses to target items. However, there are reasons to question the suitability of these measures for answering either research question. The present thesis evaluates alternative measurement models for both purposes. A meta-analysis of published and two new experiments showed that the task-oriented model, which was designed to provide pure measures of retrieval experience per se, is a valid measurement model. In two experiments, it was used to test the distinctiveness-fluency framework of retrieval experience. Although both experiments yielded the same result in terms of rememberknow response rates, the model-based analysis showed that only one manipulation affected the critical valid measure of retrieval experience. Besides casting doubt on the distinctiveness-fluency framework, this finding demonstrates empirically that remember-know response rates may lead to erroneous conclusions.