Cognitive neuropsychology has championed the use of single-case research design. Recently, however, case series designs that employ multiple single cases have been increasingly utilized to address theoretical issues using data from neuropsychological populations. In this paper, we examine these methodologies, focusing on a number of points in particular. First we discuss the use of dissociations and associations, often thought of as a defining feature of cognitive neuropsychology, and argue that they are better viewed as part of a spectrum of methods that aim to explain and predict behaviour. We also raise issues regarding case series design in particular, arguing that selection of an appropriate sample, including controlling degree of homogeneity, is critical and constrains the theoretical claims that can be made on the basis of the data. We discuss the possible interpretation of "outliers" in a case series, suggesting that while they may reflect "noise" caused by variability in performance due to factors that are not of relevance to the theoretical claims, they may also reflect the presence of patterns that are critical to test, refine, and potentially falsify our theories. The role of case series in treatment research is also raised, in light of the fact that, despite their status as gold standard, randomized controlled trials cannot provide answers to many crucial theoretical and clinical questions. Finally, we stress the importance of converging evidence: We propose that it is conclusions informed by multiple sources of evidence that are likely to best inform theory and stand the test of time.