Background: Rescue of patients (pts) who fail to be cured following 2-3 chemotherapy (CT) combinations ± high dose CT (HDCT) or who are cisplatin-refractory is still an unmet need. The identification of novel active agents/combinations is a priority. Methods: We reviewed the characteristics of consecutive pts who received TPG combination at our Institution for relapse/progression after 2 or 3 previous platinum-based CT lines. Treatment consisted of the administration of paclitaxel 80 mg/m² IV, cisplatin 50 mg/m² IV and gemcitabine 800 mg/m² IV on days 1 and 8 q3w. Program provided up to 8 administrations of TPG followed by surgery of all resectable masses. Paraffin slides of all cases with residual viable cancer after TPG were retrieved: CD30 staining and somatic mutation analysis by Sequenom OncoCarta Panel v.1.0 were planned. Results: From 04/1999 to 07/2011, 74 pts have been treated with TPG, 64 in 3rd and 10 pts in 4th line. Median age was 32 (20-55), median number of prior CDDP cycles was 7 (4-10), 10 pts (13%) had already received HDCT, 4 pts had primary mediastinal GCT, 42 (57%) were CDDP-refractory/abs refractory, 25 pts (34%) had either liver, bone or brain metastases before TPG. 40% developed G3-4 hematologic and 2 pts G3 renal toxicity, none of them being dose-limiting. No treatment delays have been observed. Median number of administrations was 7 (1-10). 8 complete responses (CR), 28 partial responses with normal markers (PRm-) and 13 incomplete response/stable disease have been recorded, for a major response rate (CR+PRm-) of 49%. 34 pts (46%) underwent post-CT surgery which was radical in 15 cases (44%) and led to 7 pathologic CR. 3-, 6- and 12-month PFS was 68% (95% CI 55-76%), 31% (21-42) and 20% (12-30), respectively. 1-yr PFS after HDCT only was 24% (4-53). 12- and 24-month OS was 46% (34-57) and 26% (16-37), respectively. Median PFS and OS were 5 (1-104) and 12 mos (1-116). Conclusions: TPG combination rates first among salvage therapies for refractory GCT, even after failure of HDCT. Benchmark survival endpoints for trials on new agents/combinations have been set. Results of the biomarker analysis will be available in Feb 2012 and may help to corroborate a window of opportunity for targeted agents in GCT.