Background: In recent years, clinical trials have shown improved survival of patients with metastatic esophageal or gastric cancer. The number of patients participating in clinical trials is limited, and survival improvements observed from clinical trials are unrepresentative for the full population. The aim of our study was to assess trends in survival for the best-case, typical, and worst-case scenarios in patients with metastatic esophageal or gastric cancer. Methods: We selected patients with metastatic esophageal or gastric cancer diagnosed between 2006 and 2020 from the nationwide Netherlands Cancer Registry. Survival was calculated for different percentiles of the survival curve for each incidence year (eg, the 10th percentile [p10] represents the top 10% of patients with the best survival): p10 (best-case), p25 (upper-typical), p50 (median), p75 (lower-typical), and p90 (worst-case). Weighted linear regression analyses were performed to test whether changes in survival were significant. Results: The overall median survival between 2006 and 2020 remained unchanged for patients with esophageal cancer (n=10,448; from 5.2 to 5.2 months, respectively; P=.06) and improved for patients with gastric cancer (n=10,512; from 3.5 to 4.3 months, respectively; P=.001). For patients with esophageal cancer, survival for the best-case scenario (p10; best 10% of patients) significantly improved from 17.2 to 21.0 months (P=.006). For patients with gastric cancer, survival significantly improved for the best-case scenario (p10) from 15.9 to 23.5 months (P<.001) and the upper-typical scenario (p25) scenario improved from 7.9 to 9.9 months (P<.001). Conclusions: Despite significant survival improvements in clinical trials, survival improvements were not observed for the majority of patients treated in daily clinical practice. An increase in survival was only observed for patients with the best prognosis.