PURPOSE: To identify demographic and disease-related characteristics predictive of LTFU status in amblyopia treatment and create a risk model for predicting LTFU status. DESIGN: Retrospective cohort study METHODS: Setting: Single center, ophthalmology department at Boston Children's Hospital (BCH). PATIENTS: 2037 patients treated for amblyopia at BCH between 2010-2014. OBSERVATION PROCEDURE: LTFU was defined as patients who did not return after initial visit, excluding those who came for second opinion. Multiple variables were tested for association with LTFU status. OUTCOME MEASURE: Odds ratio of LTFU risk associated with each variable. Multivariate logistic regression was used to create a risk score for predicting LTFU status. RESULTS: A large proportion of patients (23%) were LTFU after first visit. Older age, non-white race, lack of insurance, previous glasses or atropine treatment, and longer requested follow-up intervals were independent predictors of LTFU status. A multivariable risk score was created to predict probability of LTFU (AUC 0.68). CONCLUSIONS: Our comprehensive amblyopia database allows us to predict which patients are more likely to be LTFU after baseline visit, and develop strategies to mitigate these effects. These findings may help with practice efficiency and improve patient outcomes in the future by transitioning these analyses to an electronic medical record that could be programmed to provide continually updated decision support for individual patients based on large datasets.