In accord with classic schema theory, people are susceptible to forming false memories that align with stored schema representations (Brewer & Treyens, 1981). Furthermore, clinicians schematize mental disorders as causal networks of features (de Kwaadsteniet, Hagmayer, Krol, & Witteman, 2010; Kim & Ahn, 2002). We asked whether one important consequence of this representation is that clinicians tend to misremember client cases as being more causally coherent than they actually are. We tested this hypothesis by manipulating the causal coherence of case descriptions via a well-documented cue-to-causality, the proportionality between features (Einhorn & Hogarth, 1986). Clinicians read hypothetical cases describing three pieces of clinically relevant client information presented in causal order: recent life events, the clients' emotional reactions to those events, and their behaviors following those reactions. Each piece of information (event, reaction, behaviors) was manipulated to either be severely or mildly negative, rendering it proportionate or disproportionate to the other pieces of information. The clinicians offered diagnoses for these client cases, and then completed an unexpected recognition task. Clinicians were significantly more likely to misremember causally incoherent cases (i.e., in which the severity of the client's emotional reaction did not match the severity of the life event or behaviors) as having been coherent, compared to their likelihood of misremembering coherent cases as having been incoherent. They also incorrectly recognized false reaction lures more frequently overall than false event lures or false behavior lures. We discuss potential implications for the proportionate-response effect, schema theory, inference, causal coherence, and expert judgments.