BACKGROUND: It is unclear how lung function may recover in patients with residual lung abnormalities (RLAs) following COVID-19 pneumonia. PURPOSE: To evaluate lung function trends over time in patients with RLAs following hospitalization due to COVID-19. STUDY TYPE: Prospective, multicenter longitudinal cohort study. POPULATION: Twenty-four participants hospitalized due to COVID-19 with RLAs identified on CT ≥ 3 months postdischarge (median [IQR] age 69 (15) years; 3 female) underwent at least one MRI at 6 months (n = 16), 1 year (n = 19), or 2 years (n = 14). FIELD STRENGTH/SEQUENCE: 1.5 T. Dynamic contrast enhanced (DCE) 3D spoiled gradient echo, 129Xe steady state free precession (ventilation), 129Xe 3D spoiled gradient echo multiple b-value (diffusion-weighted), 129Xe 4-echo flyback 3D radial (dissolved phase). ASSESSMENT: Pulmonary blood flow, volume, and mean transit time (MTT) were calculated from DCE MRI. The fraction of 129Xe signal in the red blood cells to membrane (RBC:M) was calculated from the dissolved phase 129Xe acquisition. Ventilation defect percentage (VDP) was calculated from the 129Xe ventilation acquisition. Mean diffusive length scale (LmD) was calculated from the 129Xe diffusion-weighted acquisition. STATISTICAL TESTS: Changes in metrics with time and associations between metrics were assessed using mixed-effect linear regression. Correlations were tested using Spearman's correlation coefficient. Regional differences were assessed using a Friedman's test with a Bonferroni adjustment. p < 0.05 was considered significant. RESULTS: Pulmonary blood flow and MTT improved significantly over time (MTT: 6 months, 15.3 (IQR, 2.0); 1 year, 15.6 (1.4); 2 years, 15.0 (5.3); pulmonary blood flow: 6 months, 75.4 (IQR, 22.0); 1 year, 83.2 (47.4); 2 years, 107.3 (51.1)). RBC:M z-score was low at all three visits (6 months, -2.85 (0.98); 1 year, -2.44 (1.34); 2 years, -2.60 (1.39)), with no improvement with time (p = 0.993). VDP and LmD did not significantly change with time (VDP: p = 0.100; LmD: p = 0.166). DATA CONCLUSION: Improvements in lung perfusion were measured; however, there was no corresponding enhancement in RBC:M. EVIDENCE LEVEL: Level 2. TECHNICAL EFFICACY: Stage 3.