Rationale: Hyperglycemia is associated with mortality in patients with community-acquired pneumonia (CAP), and hyperglycemia may be a biomarker of severity. However, hyperglycemia has a major disadvantage because the association is diminished in patients with diabetes mellitus (DM). This hampers the use of hyperglycemia as a biomarker. Accounting for habitual glucose levels could overcome this disadvantage.Objectives: We hypothesized that the glycemic gap (the difference between plasma glucose and the estimated average glucose) may be associated with mortality irrespective of DM.Methods: Among 1,933 adults with CAP included in a prospective multicenter cohort, we investigated the association between the glycemic gap and 90-day mortality. Hemoglobin A1c was used to estimate the average glucose. The association was assessed with Cox proportional hazard models after adjustment for age, sex, CURB-65 (Confusion, urea >7 mmol/L, respiratory rate ≥30 breaths/minute, systolic blood pressure <90 mmHg or diastolic blood pressure ≤60 mmHg and age ≥65 years), and comorbidities. In the prespecified analysis the absolute and relative glycemic gaps were used as a continuous variable. In a post hoc analysis, the absolute and relative glycemic gaps were used as a categorical variable grouped according to quartiles.Results: In the post hoc analysis, patients with the lowest (negative) and highest (positive) absolute glycemic gap quartiles had increased risk of 90-day mortality (hazard ratio, 2.6; 95% confidence interval, 1.02-6.65; and hazard ratio, 2.5; 95% confidence interval, 1.01-6.06, respectively). A similar association was found for the relative glycemic gap. The associations were independent of age, CURB-65 score, sex, or number of comorbidities and not modified by DM.Conclusions: Patients with the highest and lowest glycemic gap may have an increased risk of 90-day mortality, and the association was not modified by DM. These associations were found in an exploratory post hoc analysis and should be validated in other populations before further conclusions can be made.