Background: The COVID-19 pandemic reshaped the epidemiology and outcomes of community-acquired pneumonia (CAP), yet high-quality data from low- and middle-income countries (LMIC) in Latin America remain scarce. We characterized CAP in a Colombian inpatient cohort during the pandemic, comparing COVID-19 and non-COVID-19 CAP, evaluating age-stratified risk, benchmarking three severity scores, and quantifying the prognostic impact of microbial coinfection. Methods: This retrospective analysis used the NACef (Community-Acquired Pneumonia, Endotypes, and Phenotypes) dataset of 768 adults hospitalized with CAP at Clínica Universidad de La Sabana, Colombia (January 2020–July 2022). Outcomes included ICU admission, mechanical ventilation, and in-hospital death. Continuous and categorical variables were compared with Mann–Whitney U and χ²/Fisher tests; the area under the receiver-operating-characteristic curve (AUC) benchmarked SOFA, CURB-65, and PSI; multivariable logistic regression identified independent mortality predictors. Results: Among 768 patients (mean age 59.2 ± 14.6 years; 62.8% male; 81.1% COVID-19-positive), 34.0% required ICU admission, 41.4% mechanical ventilation, and 23.6% died in hospital. COVID-19 CAP carried nearly double the mortality of non-COVID-19 CAP (26.0% vs 13.3%; relative risk 1.96, 95% CI 1.26–3.04) and far higher ventilation rates (45.9% vs 21.7%; p < 0.001), despite comparable admission severity scores. Mortality rose monotonically across age strata from 12.5% (< 50 years) to 33.9% (70–79 years), and each year of age independently increased death risk (adjusted OR 1.034/year). PSI was the strongest single discriminator of mortality (AUC 0.751) but all three scores performed similarly (DeLong p > 0.05); PSI excelled in younger patients (AUC 0.795) whereas SOFA was superior in those aged ≥ 65 (AUC 0.745). Microbial coinfection, present in 9.6% of patients, independently predicted death (adjusted OR 2.91, 95% CI 1.55–5.32). Conclusions: In this first systematic analysis of a Latin American CAP cohort spanning the pandemic, COVID-19, advancing age, and bacterial coinfection were robust, independent mortality drivers. No single severity score dominated; age- and context-tailored score selection is warranted. These findings provide locally relevant evidence for CAP management in LMIC settings.