Abstract Background In Colombia, cancer incidence is increasing, as is the demand for end-of-life care. Understanding how patients who die from cancer experience this phase will allow the identification of factors associated with greater suffering and actions to improve end-of-life care. We aimed to explore associations between the level of suffering of patients who died from cancer and were cared for in three Colombian hospitals with patient, tumor, treatment, and care characteristics and provided information. Methods Data on the last week of life and level of suffering were collected through proxies: Bereaved caregivers of patients who died from cancer in three participating Colombian hospitals. Bereaved caregivers participated in a phone interview and answered a series of questions regarding the last week of the patient’s life. An ordinal logistic regression model explored the relationship between the level of suffering reported by bereaved caregivers with the patient’s demographic and clinical characteristics, the bereaved caregivers, and the care received. Multivariate analyses were adjusted for place of death, treatments to prolong of life, prolongation of life during the dying process, suffering due to prolongation of life, type of cancer, age, if patient had partner, rural/urban residence of patient, importance of religion for the caregiver, caregivers´ relationship with the patient, and co-living with the patient. Results A total of 174 interviews were included. Median age of the deceased patients was 64 years (IQR 52–72 years), and 93 patients were women (53.4%). Most caregivers had rated the level of suffering of their relative as “moderately to extremely” (n = 139, 80%). In multivariate analyses, factors associated with a higher level of suffering were: unclear information about the treatment and the process before death Odds Ratio (OR) 2.26 (90% CI 1.21–4.19), outpatient palliative care versus home care OR 3.05 (90% CI 1.05–8.88), procedures inconsistent with the patient’s wishes OR 2.92 (90% CI 1.28–6.70), and a younger age (18–44 years) at death versus the oldest age group (75–93 years) OR 3.80 (90% CI 1.33–10.84, p = 0.04). Conclusion End-of-life care for cancer patients should be aligned as much as possible with patients´ wishes, needs, and capacities. A better dialogue between doctors, family members, and patients is necessary to achieve this.