Background: There are few reports on the epidemiology of gastric cancer metastasis, although outcomes are known to be uniformly poor. Here we describe the patterns of gastric cancer metastasis and treatment in the United States (US). Methods: Patients with gastric adenocarcinoma histologies were identified in the National Cancer Database (NCDB) from 2004-2016. We describe univariate associations between different sites of metastasis and clinicopathologic characteristics and treatment modalities, using the chi-square and Kruskal-Wallis tests. Kaplan Meier curves were constructed for the estimation of overall survival (OS) by metastatic site. Results: Due to changes in the coding of metastatic disease, we were limited to the year 2016 for evaluation of patterns of disease. Twenty-six percent (n = 1228) of gastric cancer patients presented with liver metastases, 20% (n = 941) with distant nodes, 43% (n = 2028) with other distant site metastases (including peritoneum), and the rest to bone, brain or lung. On univariate analysis, when compared to liver metastases, other distant site metastases were significantly more likely to arise from an antral primary site (28% v. 16%); to be of Hispanic origin (16% v. 7%); female (42% v. 29%); associated with signet ring histology (34% v. 6%); lymphovascular invasion (LVI) (58% v. 27%); and tumor grade III/IV (85% vs. 60%) (p < 0.0001 for all). There were no significant differences in how patients with metastatic disease were treated in terms of systemic therapy. With regard to OS, due to how metastatic sites were coded prior to 2016, it was not possible to compare peritoneal metastases against other sites; however, patients with distant nodal disease had improved median overall survival compared to those with any other metastatic site (7.9 v. 5.2 months, p < 0.0001). Conclusions: The majority of US patients with metastatic gastric cancer present with presumed peritoneal disease. Predictive factors for peritoneal metastases vs. liver metastases included adverse prognostic features, including signet ring histology, higher tumor grades, and LVI. Although it was not possible to compare OS of peritoneal disease against other sites, continued follow up is needed, as this may impact future staging.