Genotype × environment interactions (GEIs) should play an important role in the selection of suitable germplasm in breeding programmes. We here assessed GEI effects on pearl millet (Pennisetum glaucum L.) genotypes, selected to possess a high concentration of slowly digestible starch (SDS) and resistant starch (RS) in their grains. Entries were grown in a randomized complete block design with three replications at locations in Bawku-Ghana, Sadore-Niger, Bamako-Mali, Konni-Nigeria, and Gampella-Burkina Faso across West Africa. Harvested grains from these locations were metabolomically profiled using flow injection ionization-high-resolution mass spectrometry (FIE-HRMS). A total of 3144 mass features (m/z) (1560 negative ion mode and 1584 positive ion mode) were detected, of which, 475 m/z were linked to metabolites be involved in starch, antioxidant and lipid biosynthesis, and vitamin metabolism. Combined ANOVA revealed that the GEI was significantly evident for 54 health-benefiting metabolites, many associated with sugar, especially galactose, metabolism. Additive main effects and multiplicative interaction (AMMI) analysis examined genotype variation and GEI effects, which, when combined with principal component analysis (PCA), found that m/z 171.14864 (positive ionisation, propenyl heptanoate) accounted for 89% of the GEI variation along PC1. The AMMI-based stability parameter (ASTAB), modified AMMI stability value (MASV), and modified AMMI stability index (MASI) were then applied to identify stable and high-performing genotypes for all the health-benefiting metabolites. Similarly, the best-linear-unbiased-prediction (BLUP)-based stability estimation was also performed using the harmonic mean of genotypic values (HMGV), relative performance of genotypic values (RPGV), and harmonic mean of relative performance of genotypic values (HMRPGV), to identify genotype rankings across multiple environments. The multi-trait stability index (MTSI) was calculated and found that the genotypes G1 (ICMH-177111) and G24 (ICMX-207137) were the most stable and were the best mean performers across 52 health-benefiting metabolic traits. These findings demonstrate the potential of G × E assessments on the delivery of health-benefiting metabolite-rich grains in future varieties and hybrids of pearl millet.