Palliative care research is challenged by a disagreement as to what palliative care is, when it should be offered and what conditions warrant specialized palliative care services. These challenges became evident when we used a population-based data linkage to evaluate the delivery of palliative care services in Western Australia. This paper describes the development of a conceptual framework to provide minimal, mid-range, and maximal estimates of a palliative care population. The estimates include nonmalignant conditions; realistically restrict the number and types of conditions; and propose a time frame over which specialized services can be offered. In defining a palliative care population for the purpose of research, development of an estimation method simultaneously addressed the rhetoric of palliative care with the restrictions inherent in both population-based research and service delivery. When applied to a population, the 10 conditions of the minimal estimate provide an acceptable guide to future research and practice.