OBJECTIVE: To model the global test-retest variability of visual fields (VFs) in glaucoma. DESIGN: Retrospective cohort study PARTICIPANTS: 8,088 VFs of 4,044 eyes from 4,044 participants. METHODS: We selected two reliable VFs (SITA 24-2) per eye measured with the Humphrey Field Analyzer within 30 days of each other. Each VF contained fixation losses (FL) ≤ 33%, false-negative rates (FNR) ≤ 20%, and false-positive rates (FPR) ≤ 20%. Stepwise linear regression was applied to select the model that best predicts the global test-retest variability from three categories of features of the first VF: (1) base parameters (age, mean deviation [MD], pattern standard deviation, glaucoma hemifield test, FPR, FNR, FL); (2) total deviation (TD) at each location; and (3) computationally-derived VF loss patterns (archetypes). The global test-retest variability was defined as root mean square deviation (RMSD) of TD values at all 52 VF locations. Model performance was assessed using adjusted R-squared and Bayesian information criterion (BIC). MAIN OUTCOME MEASURES: Archetype models to predict the global test-retest variability. RESULTS: The mean ± standard deviation of RMSD was 4.39 ± 2.55 dB. Between the two VF tests, TD values were more strongly correlated in central than in peripheral VF locations (intraclass coefficient range: 0.66-0.89; p < 0.001). Compared with the model using base parameters alone (adjusted R-squared = 0.45), adding TD values improved prediction accuracy of the global variability (adjusted R-squared = 0.53, p < 0.001) and BIC (decreased by 527; a change of > 6 represents strong improvement). Lower TD sensitivity in the outer-most peripheral VF locations was predictive of higher global variability. Adding archetypes to the base model improved model performance with an adjusted R-squared of 0.53 (p < 0.001) and lowering of BIC by 583. Greater variability was associated with concentric peripheral defect, temporal hemianopia, inferotemporal defect, near total loss, superior peripheral defect, and central scotoma (listed in order of decreasing statistical significance), and less normal VF and superior paracentral defect. CONCLUSIONS: Inclusion of archetype VF loss patterns and TD values based on first VFs improved the prediction of the global test-retest variability than using traditional global VF indices alone.