Elderly patients may be heterogeneous in their abilities to tolerate immunochemotherapy-associated toxicities. We describe the morbidity of rituximab-chemotherapy combinations among 205 newly-diagnosed diffuse large B-cell lymphoma (DLBCL) patients aged ≥60 years from 3 tertiary hospitals between 2009 and 2016, and explore the utility of retrospectively-assigned baseline Comprehensive Geriatric Assessment (CGA) in predicting these toxicities. Seventy-three percent (146/201) experienced grade ≥3 toxicities, 81% (163/201) needed admission, 52% (107/205) had ≥2 unplanned admissions, 82/201 (41%) required dose reductions (DR) subsequent to Cycle 1, 39/166 (23%) had chemotherapy delays and 26/198 (13%) ceased therapy early. CGA was associated with pre-emptive baseline DR and perhaps because of this, did not predict grade ≥3 toxicities, ≥2 unplanned admissions or subsequent DR. Three-year overall survival (OS) of CGA-fit, CGA-unfit and CGA-frail patients was 82%, 60% and 53%, respectively. Three-year progression-free survival (PFS) of CGA-fit, CGA-unfit and CGA-frail patients was 66%, 58% and 46%, respectively. OS of CGA-fit patients was not statistically different from CGA-unfit patients, but was superior to CGA-frail patients (hazard ratio 2·892, 95% confidence interval 1·275–6·559, P = 0·011). PFS differences were not statistically significant. Baseline DR and early therapy cessation were associated with inferior OS and PFS independent of CGA. Prospective studies are needed to confirm if CGA-adapted treatment strategies minimize morbidity and improves survival.