Plant breeders have made significant progress in developing high-yielding cultivars. They have used available working collections consisting of elite lines and varieties in achieving their short-term objectives of developing cultivars. The large genetic variability available in the ex-situ collections has not been adequately used in most crop improvement programs. Even at ICRISAT, that houses the world peanut germplasm collection of > 14 000 lines, less than 1% were used in developing 8279 breeding lines from 1986 to 2002. Thus most cultivars have a narrow genetic base. This exiguous use of germplasm is due to lack of reliable information on traits of economic importance, which show high genotype x environment interactions and require replicated multilocational evaluations to identify useful parents. Development of core (10% of entire collection) and mini-core (10% of core or 1% of entire collection) in case of large germplasm collections has been suggested as a means to enhance the use germplasm. We have developed core collections of five ICRISAT mandate crops: sorghum (2247 accessions), pearl millet (1600), chickpea (1956), pigeonpea (1290), and peanut (1704) and finger millet (622); and mini-core collections in chickpea (211 accessions, entire collection >17, 000) and peanut (184 accessions, entire collection > 14, 000). Using core and mini-core approaches we have identified diverse sources of early-maturity, high yield, tolerance to drought and low temperature at germination, oil and protein contents in peanut; high yield in pigeonpea; and for early-maturity, drought tolerance, high yield, and large seed size in kabuli chickpea. The molecular characterization of core and mini-core collections and trait specific germplasm will result in identifying genetically diverse parents and efficient gene mining. Their inclusion in working collections will add value to the working collections and use in crop improvement will result in enhancement of the trait and broaden the genetic base of cultivars.