Population-based studies investigating access to palliative care often use death in a hospice as a proxy for service use. We linked data from a large South London hospice to Thames Cancer Registry (TCR) data to determine whether patients who received hospice services differed from those who did not. We matched hospice data for 2474 cancer patients dying between 2000 and 2006, while resident within a restricted catchment area, to TCR data for residents in this area. During matching 14.2% (n = 352) of hospice patients were excluded due to differing key dates or addresses. In addition, 5.6% (n= 175) of residents initially defined as not receiving hospice services were recorded as dying in a hospice in the TCR dataset. The problems of overlapping catchment areas and of defining patients receiving services meant we could not adequately determine use of hospice services. This method might be applied more successfully to non-urban hospices, primary care trusts or larger regions.