Surveillance frequency for metastasis is guided by gene expression profiling (GEP). This study evaluated the effect of GEP on time to diagnosis of metastasis, subsequent treatment and survival. A retrospective study was conducted of 110 uveal melanoma patients with GEP (DecisionDx-UM, Castle Biosciences, Friendswood, Texas, USA) and 110 American Joint Committee on Cancer-matched controls. Surveillance testing and treatment for metastasis were compared between the two groups and by GEP class. Rates of metastasis, overall survival and melanoma-related mortality were calculated using Kaplan–Meier estimates. Baseline characteristics and follow-up time were balanced in the two groups. Patients’ GEP classification was 1A in 41%, 1B in 25.5% and 2 in 33.6%. Metastasis was diagnosed in 26.4% (n = 29) in the GEP group and 23.6% (n = 26) in the no GEP group (P = 0.75). Median time to metastasis was 30.5 and 22.3 months in the GEP and no GEP groups, respectively (P = 0.44). Median months to metastasis were 34.7, 75.8 and 26.1 in class 1A, 1B and 2 patients, respectively (P = 0.28). Disease-specific 5-year survival rates were 89.4% [95% confidence interval (CI): 81.0–94.2%] and 84.1% (95% CI: 74.9–90.1%) in the GEP and no GEP groups respectively (P = 0.49). Median time to death from metastasis was 10.1 months in the GEP group and 8.5 months in the no GEP group (P = 0.40). There were no significant differences in time to metastasis diagnosis and survival outcomes in patients with and without GEP. To realize the full benefit of GEP, more sensitive techniques for detection of metastasis and adjuvant therapies are required.