Objective: The effectiveness of inpatient palliative care units, a complex intervention, is challenging to evaluate due to methodological and practical difficulties. We conducted a study to evaluate providers' perceived effectiveness of one such unit. Methods: A non-concurrent, prospective, controlled study using the Audit Scale for good death services as an indicator of process of care and the Good Death Scale as the outcome of provider assessment of quality of dying was conducted. Eighty of 212 terminally ill cancer patients were matched from a tertiary medical center in Taiwan. Patients in the unit served as the intervention group and patients in the oncology ward served as controls. Multiple logistic regression was applied to estimate the propensity of choosing the unit for each patient, and linear regression analysis was conducted to identify predictive factors for mean change scores of the Good Death Scale. Results: Male gender (P , 0.001, 95% confidence interval ¼ 0.73-2.43) was associated with better quality of dying while having hepatocellular carcinoma (P , 0.004, 95% confidence interval ¼ 22.22 to 20.44) was associated with worse quality. For those in the unit, higher total Audit Scale scores were positively related to the outcome of quality of dying. The unit (P , 0.001, 95% confidence interval ¼ 8.67-12.97) and higher Good Death Scale at admission (P , 0.001, 95% confidence interval ¼ 0.44-1.13) were predictors of Audit Scale scores. Conclusions: Admission to a palliative care unit was associated with higher provider assessments of quality of dying for terminally ill cancer patients. These units should be considered as options for hospitals looking for ways to improve the quality of dying for patients.