Knowledge of the long term trajectory of nonadherence to immunosupressants can inform decisions regarding organ allocation, adherence monitoring, and intervention efforts. The MALT (Medication Adherence in children who had a Liver Transplant) prospective multi-site study followed 400 pediatric and adolescent liver transplant recipients for 2 years, using the Medication Level Variability Index (MLVI) to monitor adherence. We hypothesized that adherence is an unstable (fluctuating) phenomenon: that patients who are adherent in year one may become nonadherent in year 2, and vice-versa. However, we also hypothesized that a majority (more than 50%) of nonadherent patients remain nonadherent over time. We further hypothesized that the longer nonadherence lasts, the higher the likelihood of adverse events (rejection). Finally, we explored the effect of socioeconomic factors on the evolution of adherence over time. Most (59.7%) of MALT patients who were nonadherent in year 1 remained so in year 2; 18.5% of patients who were adherent in year 1 became nonadherent in year 2. Only 4.4% of patients who were adherent in both year 1 and year 2 had a rejection, compared with 22.9% of patients who were nonadherent during one of the years, and 34.9% of those who were nonadherent in both years (p<.001), establishing a "dose dependent" effect of adherence on transplant outcomes. Single-parent households were associated with worsening adherence. Our results suggest that good baseline adherence does not guarantee adherence later on, that nonadherence is likely to persist in the absence of interventions, and that monitoring of adherence and interventions to improve it should be expected to last for years if transplant outcomes are to be improved. This article is protected by copyright. All rights reserved.