OBJECTIVE: Determine if pre-treatment Patient-Generated Subjective Global Assessment (PG-SGA) predicts febrile neutropenia (FN) in gynecologic cancer patients receiving primary combination chemotherapy. METHODS: Following IRB approval, clinicopathologic variables, pre-treatment laboratory values and PG-SGA were recorded from eligible patients. Bone marrow toxicity (CTC 3.0) divided groups of patients: (1) No grade 3 or 4 neutropenia, (2) grade 3 or 4 neutropenia, (3) FN. Statistical analysis with Kruskal-Wallis one-way analysis of variance and a receiver operating characteristic (ROC) curve were performed. RESULTS: 58 patients met study inclusion: 25 in group 1, 28 in group 2, and 5 in group 3. Mean age was 61 and the majority, 42 (72%), had ovarian cancer. Median PG-SGA scores were: 6 (group 1) vs. 7 (group 2) vs. 14 (group 3) (p=0.019). Both median albumin: (1) 4.2 vs. (2) 4.0 vs. (3) 3.4 g/dl (p=0.041), and hemoglobin: (1) 12.1 vs. (2) 11.75 vs. (3) 10.6g/dl (p=0.05) differed between the groups. The overall AUC of the ROC curve for PG-SGA was 0.831 ± 0.064 (95% CI=0.706 to 0.956, p=0.015). Using the ROC, selecting a PG-SGA score of 7.5 to be predictive of febrile neutropenia yields a sensitivity of 100% and a specificity of 60%. When the cutoff value is set at 12.5, the specificity improves to 81% while decreasing sensitivity to 80%. CONCLUSIONS: PG-SGA scores were higher for patients experiencing FN and may be a reasonably predictive marker of FN in patients receiving multi-agent primary chemotherapy and likely benefactors of prophylactic GCSF.