159 Background: Giving chemotherapy (CTX) near the end of life is futile and costly in human and financial terms. We analyzed patterns of chemotherapy administered at the end of life (CAEOL) by disease, provider, and line of therapy as part of a broad quality initiative. We hypothesized that measuring CAEOL in our practice could identify modifiable practice patterns. Methods: All patients (pts) with an ICD-9 diagnosis of cancer who died between 1/09 and 6/10 were identified by EMR or by searching SSDI records. We focused on the cohort receiving CTX within 14 or 28 days (d) of death, stratified by tumor type, provider and line of therapy by electronic chart review. Initial results were distributed to practice physicians for review and discussion with a clinic-wide goal to be below 5% and 10% at 14 and 28d, respectively. Subsequent cohorts were then evaluated periodically with results disseminated. Results: There was a 14.5% reduction in CAEOL within 14d and a 25% reduction within 28d over time (Table), meeting the clinic-wide standard. In the 2009-2010 cohort there was a difference in 1st line pts receiving CTX within 28 d for NSCLC (24/56 pts, 42.9%) or pancreas cancer (13/19, 68.4%) and the next 4 most common diagnoses: breast, colorectal, small cell lung, and head and neck cancer (15/84, 17.9%), p= .0024. CAEOL beyond 3rd line in non-breast, or >5th line in breast cancer was uncommon. In the 2013 cohort, 1st line accounted for 28/51 pts (54.9%) receiving CTX within 28 d with NSCLC (8/15), pancreas (3/3) and lymphoma (4/4) the most prevalent diagnoses. Conclusions: CAEOL is a relatively rare event in our practice, now meeting practice standards after measurement and delving into root causes. A certain irreducible number of pts will be treated at the end-of-life, including those with potentially life-prolonging intent. First line therapy represents the majority, so identifying inappropriate candidates at diagnosis remains a quality improvement opportunity. [Table: see text]