Cultivated groundnut (Arachis hypogaea L.) is an allotetraploid oilseed legume valued for its high seed oil (48–50%) and protein (24–29%) content, yet a narrow genetic base limits breeding progress under changing climates. This study assessed phenotypic variability and genetic divergence among 34 groundnut genotypes to identify superior donors and guide hybridization. A field trial was conducted in a Randomized Block Design with three replications during Kharif 2025 at Dr. Kalam Agricultural College, Kishanganj, Bihar. Ten traits, including morphological/yield components and seed protein content (Lowry’s method), were recorded. Divergence was quantified using Mahalanobis’ D² statistics and Tocher’s clustering. ANOVA showed highly significant genotypic differences (p < 0.05) for all traits. BAU 13 excelled in pod number, pod weight, 100-kernel weight, and kernel yield; GJG 9 had the highest protein content (29.03%); and KAUSHAL combined early maturity, short stature, and high branching. Tocher’s method formed five clusters, with maximum inter-cluster distance between ICGS 44 and KAUSHAL. Divergence was driven mainly by kernels per pod, protein content, and plant height. Crossing ICGS 44 with KAUSHAL or high-yielding Cluster II accessions offers a promising route to exploit phenotypic divergence in breeding programs.