Rationale: Cross-sectional analysis of mucus plugs in computed tomography (CT) lung scans in the Severe Asthma Research Program (SARP)-3 showed a high mucus plug phenotype. Objectives: To determine if mucus plugs are a persistent asthma phenotype and if changes in mucus plugs over time associate with changes in lung function. Methods: In a longitudinal analysis of baseline and Year 3 CT lung scans in SARP-3 participants, radiologists generated mucus plug scores to assess mucus plug persistence over time. Changes in mucus plug score were analyzed in relation to changes in lung function and CT air trapping measures. Measurements and Main Results: In 164 participants, the mean (range) mucus plug score was similar at baseline and Year 3 (3.4 [0-20] vs. 3.8 [0-20]). Participants and bronchopulmonary segments with a baseline plug were more likely to have plugs at Year 3 than those without baseline plugs (risk ratio, 2.8; 95% confidence interval [CI], 2.0-4.1; P < 0.001; and risk ratio, 5.0; 95% CI, 4.5-5.6; P < 0.001, respectively). The change in mucus plug score from baseline to Year 3 was significantly negatively correlated with change in FEV1% predicted (rp = -0.35; P < 0.001) and with changes in CT air trapping measures (all P values < 0.05). Conclusions: Mucus plugs identify a persistent asthma phenotype, and susceptibility to mucus plugs occurs at the subject and the bronchopulmonary segment level. The association between change in mucus plug score and change in airflow over time supports a causal role for mucus plugs in mechanisms of airflow obstruction in asthma.