Background: Dysglycemia (i.e., prediabetes or diabetes) in patients with tuberculosis (PWTB) is associated with increased odds of mortality and treatment failure. Whether such association holds true when dysglycemia is transient or persistent is unknown. In this study, we tested the association between persistent dysglycemia (PD) during anti-tuberculosis (TB) treatment and unfavorable treatment outcomes in PWTB from Lima, Peru. Methods: PWTB enrolled between February and November 2017 were followed for 24-months. Dysglycemia was measured through fasting glucose and HbA1c at baseline during the 2nd- and 6th-month of TB treatment. PD was defined as dysglycemia detected in 2 different visits. The association between PD and unfavorable TB treatment outcome was evaluated using logistic regression. Results: Among 125 PWTB, PD prevalence was 29.6%. PD was associated with more lung lesion types, higher bacillary loads, low hemoglobin (Hb), and high body mass index (BMI). Unfavorable TB treatment outcome was associated with older age, higher BMI, more lung lesion types, and PD. After adjusting for age, Hb levels, smoking, and smear grade, PD was independently associated with unfavorable treatment outcomes (adjusted odds ratio (aOR): 6.1; 95% CI: 1.9–19.6). Conclusion: PD is significantly associated with higher odds of unfavorable TB treatment outcomes. Dysglycemia control during anti-TB treatment gives the opportunity to introduce appropriate interventions to TB management.