PURPOSE: Incidence studies reported more end-of-life decisions with possible/certain life-shortening effect (ELDs) among cancer patients than among noncancer patients. These studies did not correct for the different proportions of sudden/unexpected deaths of cancer versus noncancer patients, which could have biased the results. We investigated incidences and characteristics of ELDs among nonsudden cancer and noncancer deaths. METHODS: We sampled 5,005 certificates of all deaths in 2001 (Flanders, Belgium) stratified for ELD likelihood. Questionnaires were mailed to the certifying physicians. Data were corrected for stratification and nonresponse. RESULTS: The response rate was 59%. Among 2,128 nonsudden deaths included, ELDs occurred in 74% of cancer versus 50% of noncancer patients (P < .001). Symptom alleviation with possible life-shortening effect occurred more frequently among cancer patients (P < .001); nontreatment decisions occurred less frequently (P < .001). The higher incidence of lethal drug use among cancer patients did not hold after correcting for patient age. Half of the cancer patients who died after an ELD were incompetent to make decisions compared with 76% of noncancer patients (P < .001). Discussion with patients and relatives was similar in both groups. In one fifth of all patients the ELD was not discussed. CONCLUSION: ELDs are common in nonsudden deaths. The different incidences for symptom alleviation with possible life-shortening effect and nontreatment decisions among cancer versus noncancer patients may be related to differences in dying trajectories and in timely recognition of patient needs. The end-of-life decision-making process is similar for both groups: consultation of patients and relatives can be improved in a significant minority of patients.