Abstract Background Sepsis and pneumonia in the elderly comprise a significant portion of medical admissions. Chloramphenicol has been used in Israel for treatment of bacterial infections, without evidence regarding its efficacy and safety. Objectives We aimed to examine whether chloramphenicol was associated with similar outcomes to ceftriaxone, for treatment of sepsis and pneumonia in the elderly with dementia and functional disability. Methods Patients over 75, with dementia and functional disability, admitted to the internal medicine ward at Beilinson Hospital between 2011 and 2021, with community-acquired aspiration pneumonia or sepsis of undetermined source were included. Patients with mild dementia and independent in their activities of daily living were excluded. Primary outcome was 30- and 90-day all-cause mortality. A propensity-weighted multivariable model was constructed using inverse probability of treatment weighting. Results were expressed as OR with 95% CI. Results In total, 1558 patients were included: 512 treated with chloramphenicol and 1046 with ceftriaxone. The cohort consisted of elderly patients (mean age 87 ± 6.2 years) with comorbidities; 30- and 90-day all-cause mortality were similar [222/512 (43.3%) versus 439/1046 (41.9%) P = 0.602, and 261/512 (50.9%) versus 556/1046 (53.1%) P = 0.419, respectively]. Propensity-weighted, logistic multivariable analysis for 30- and 90-day all-cause mortality revealed similar mortality rates for chloramphenicol and ceftriaxone (OR 1.049 95% CI 0.217–1.158, OR 0.923 95% CI 0.734–1.112, respectively). Conclusion In this retrospective cohort of elderly debilitated patients hospitalized with pneumonia and sepsis, we found no difference in 30- and 90-day mortality between those treated with chloramphenicol or ceftriaxone. Further studies should determine the efficacy and safety of chloramphenicol in this population.