Abstract Limitations in functional status, frailty, multiple comorbidities and comedications are common among older colorectal cancer (CRC) patients. We investigated whether adding these factors could improve the predictive value of a reference model containing age, sex, tumor stage and location for prediction of 5‐year overall survival (OS), disease‐free survival (DFS), disease‐specific survival (DSS), recurrence‐free survival (RFS) and nondisease‐specific survival (nDSS) for all CRC patients as well as for younger (<65 years) and older patients (≥65 years). Overall, 3410 CRC patients from the DACHS study were analyzed and area under receiver operating characteristic curves (AUC) and net reclassification improvements (NRI) were assessed. In prediction of OS, the reference model plus functional status was identified as the best model among all CRC patients (AUC: 0.762) and younger CRC patients (AUC: 0.820). In older CRC patients, comorbidity should additionally be added (AUC: 0.747). For nDSS, the reference model plus comorbidity and frailty had the best predictive performance in all CRC patients (AUC: 0.776). For the outcomes DFS (AUC: 0.727), DSS (AUC: 0.838) and RFS (AUC: 0.784), the reference model was already the best model in all CRC patients because no significant NRIs were observed. The pattern “The less CRC‐specific the survival outcome and the older the CRC patients, the more relevant the inclusion of functional status, comorbidity, and frailty in CRC prognostic scores is” was observed. Thus, different nomograms for younger and older CRC patients for 1‐, 3‐ and 5‐year OS prognosis estimation are being suggested.