Ecological Momentary Assessment (EMA) is a widely used methodology to examine psychological and behavioral phenomena among individuals with eating disorders (EDs). While EMA overcomes limitations associated with traditional retrospective self-report, it remains subject to potential methodological limitations, including poor adherence to the EMA protocol, which may bias findings. Little is known about baseline and momentary predictors of missing EMA data in ED research; however, such work may help clarify the correlates of missingness and illuminate steps to address potential bias. The purpose of this study was to investigate predictors of EMA adherence in a sample of adults with binge-eating disorder (BED) enrolled in a randomized treatment trial. Prior to treatment, 110 patients completed self-report questionnaires assessing demographics, psychopathology, and transdiagnostic risk/maintenance factors. Participants then responded to EMA questions regarding their eating behavior and internal states six times a day for seven days. A series of generalized-linear and mixed-effect models were conducted to examine baseline and momentary predictors of EMA adherence. No significant baseline predictors were identified, suggesting that participants' overall level of missing data was not related to person-level characteristics (e.g., gender, level of ED pathology). However, lower positive affect, lower hunger, signals later in the day, later days in the EMA protocol, and missed prior signals predicted greater odds of signal non-response, suggesting certain contextual factors may impact the likelihood that a participant with BED will respond to the subsequent EMA signal. Ultimately, these findings have implications for future eating disorder EMA research.