The objectives of this study were (i) to compare the effectiveness of a desired gain index, direct selection, and an estimated index for improvement of corn (Zea mays L.) grain yield and percent protein and (ii) to compare the effectiveness of mass selection and half‐sib family selection for increasing percent protein. Seven selection method‐index combinations were applied to two corn populations (RSSSC, a composite of four stiff‐stalk selections and RSL, a ‘Lancaster’ population) in each of 2 years. For half‐sib family selection, 200 families from each population were evaluated in three replications each year. For mass selection, 500 ears, 25 ears from each of 20 blocks, were analyzed for protein and kernel weight.Selected progenies were evaluated in 20 replications in 1977. Indices compared were: (i) desired gain indices to (a) increase yield 20% and jprotein to 13% (HSPY) and (b) increase percent protean while holding kernel weight constant using half‐sib family means (HSPK) mass selection (MSPK); (ii) direct selection for grain yield (HSBY) or percent protein (HSBP); (iii) estimated indices to maximize gain in protein (HSEP) or grain yield (HSEY) using information on both traits.Mass selection was as effective as half‐slb family selection for increasing percent protein; simultaneous improvement of grain yield and percent protein using a desired gain index was feasible; agreement with prediction of gains from an estimated index or a desired gain index was as good as for single‐trait selection; genetic variance and covariance estimates from 200 half‐sib families were sufficiently accurate to be useful in index development.