Electronic diary data, such as that acquired through Ecological Momentary Assessments (EMA), has historically provided novel insights into diverse psychological processes. Analyses of these data typically focus on modeling participant-specific means, variability, and stability. We propose a novel statistical framework to determine participant stability by quantifying fragmentation of standardized trajectories using the following 2-step approach: (1) participant-level EMA scores are normalized, and (2) normalized scores are dichotomized into 2 states, inside and outside a range of 1 standard deviation. Within-participant fragmentation measures were calculated from dichotomized scores and modeled with various covariates. We used this method to study patterns of emotional states and showed that the proposed fragmentation measures differentiate mood disorder subtypes, including Bipolar I (BPI), Bipolar II, and major depressive disorder (MDD) compared with unaffected controls. Fragmentation measures were regressed on the mood disorder subtype, adjusting for age, sex, body mass index, and mean squared successive difference. The analyses revealed decreased stability (more fragmentation) among those with BPI when inside the participant-specific standard range of attention (β = 0.09, p = .004) and decreased stability among those with MDD inside the standard range of mood (β = 0.04, p = .039) and attention (β = 0.05, p = .017). This work provides an illustration of the clinical significance of EMA in characterizing the stability of mood, attention, or other psychological states that may underlie psychological disorders and phenomena. The application of fragmentation provides a novel statistical approach that can characterize within-participant stability beyond currently available traditional approaches. (PsycINFO Database Record (c) 2019 APA, all rights reserved).