Pigeonpea, a climate-resilient legume, is nutritionally rich and of great value in Asia, Africa, and Caribbean regions to alleviate malnutrition. Assessing the grain nutrient variability in genebank collections can identify potential sources for biofortification. This study aimed to assess the genetic variability for grain nutrients in a set of 600 pigeonpea germplasms conserved at the RS Paroda Genebank, ICRISAT, India. The field trials conducted during the 2019 and 2020 rainy seasons in augmented design with four checks revealed significant differences among genotypes for all the agronomic traits and grain nutrients studied. The germplasm had a wider variation for agronomic traits like days to 50% flowering (67-166 days), days to maturity (112-213 days), 100-seed weight (1.69-22.17 g), and grain yield per plant (16.54-57.93 g). A good variability was observed for grain nutrients, namely, protein (23.35-29.50%), P (0.36-0.50%), K (1.43-1.63%), Ca (1,042.36-2,099.76 mg/kg), Mg (1,311.01-1,865.65 mg/kg), Fe (29.23-40.98 mg/kg), Zn (24.14-35.68 mg/kg), Mn (8.56-14.01 mg/kg), and Cu (7.72-14.20 mg/kg). The germplasm from the Asian region varied widely for grain nutrients, and the ones from African region had high nutrient density. The significant genotype x environment interaction for most of the grain nutrients (except for P, K, and Ca) indicated the sensitivity of nutrient accumulation to the environment. Days to 50% flowering and days to maturity had significant negative correlation with most of the grain nutrients, while grain yield per plant had significant positive correlation with protein and magnesium, which can benefit simultaneous improvement of agronomic traits with grain nutrients. Clustering of germplasms based on Ward.D2 clustering algorithm revealed the co-clustering of germplasm from different regions. The identified top 10 nutrient-specific and 15 multi-nutrient dense landraces can serve as promising sources for the development of biofortified lines in a superior agronomic background with a broad genetic base to fit the drylands. Furthermore, the large phenotypic data generated in this study can serve as a raw material for conducting SNP/haplotype-based GWAS to identify genetic variants that can accelerate genetic gains in grain nutrient improvement.