Background: Interplay between non-specific inflammatory reaction and tumor microenvironment in gastric cancer (GC) can be measured indirectly by assessing fluctuations in concentration of platelets. Cytotoxic chemotherapy affects these morphotic elements directly by inducing myelosuppression. It was hypothesized that chemotherapy not only directly affects malignant cells, but also through immunomodulation related to myelosuppression. Methods: Metastatic GC patients (N: 155) treated with chemotherapy +/− trastuzumab were enrolled in this retrospective study. Platelet pretreatment concentration (PLT-count) and the deepest level of platelet reduction, as well as other inflammatory and general confounders were collected in the first 12 weeks of treatment (PLT-red). Martingale residuals were used to visualize the relationship between PLT-count, PLT-red, and overall survival (OS). Multiple multivariate Cox regression models were built to assess the impact of platelet reduction on OS and progression-free survival (PFS). Results: Reduction of PLT (PLT-red) to 60% of baseline concentration was associated with improved survival rates (HR = 0.60, p = 0.026 for OS and HR 0.56, p = 0.015 for PFS). Cross-classification into four groups based on PLT-count (high vs low) and PLT-red (high vs low) showed significantly worse survival rates in both high PLT-count (HR = 3.60, p = 0.007 for OS and HR = 2.97, p = 0.024 for PFS) and low PLT-count (HR = 1.75, p = 0.035 for OS and HR = 1.80, p = 0.028 for PFS) patients with insufficient platelets reduction. Conclusion: Thrombocytosis reduction represents a novel, clinically important, prognostic factor for OS and PFS in patients with stage IV GC.