Background: Sarcopenia is a poor prognostic factor in several cancers. This study aimed to assess the predictive value of sarcopenia on survival outcomes among pts with mGC. Methods: We retrospectively analyzed 170 consecutive pts with mGC starting first-line chemotherapy at the single institution during January 2013–December 2014. Inclusion criteria were as follows: treatment with fluoropyrimidine plus platinum-based regimen, ECOG PS 0–2, and adequate organ function. Skeletal muscle index (SMI) was calculated from cross-sectional area of muscle at L3 using CT scan divided by the square of the height. Baseline sarcopenia (S) was defined as 53.4 cm²/m² for men and 38.5 cm²/m²for women. Muscle loss after chemotherapy (L) was defined as 10% ≤ reduction in SMI. Univariate and multivariate analyses were conducted to identify whether S and L were independent prognostic factors of time to treatment failure (TTF) and overall survival (OS). Multivariate analysis included variables with p values ≤ 0.1 in univariate analyses. Results: Ninety-seven pts met the inclusion criteria; 83 (86%) were classified with S, and 39 (31%) with L. PS (0 vs. 1 and 2) and ALP (ULN> vs. ULN≤) were significantly associated with L. Overall response rate for S, non-S, L, and non-L was 60%, 63%, 63%, and 61%, respectively. S was not associated with TTF or OS. However, L was significantly associated with a shorter TTF (4.7 vs. 7.3 m, HR=1.82; 95% CI: 1.16-2.85, p=0.009) and tended toward worse OS (14.2 vs. 19.7 m, HR=1.31; 95% CI: 0.78-2.20, p=0.29). Variables with p values ≤ 0.1 in univariate analyses for TTF were L, LDH (ULN≤), ALB (ULN>), and histology (poorly differentiated). Histology was only variables with p values ≤ 0.1 for OS (p=0.002). As per multivariate analyses, only L was a significant prognostic factor of TTF. Conclusions: L is an independent negative prognostic factor of TTF in pts with mGC. Intervention for muscle loss can be a therapeutic approach for mGC.