AbstractThe definition of efficient selection methods for biomass yield of alfalfa and other open‐pollinated forage crops has lagged behind, despite its crucial importance for breeders. This study, performed in northern Italy, aimed to compare nine breeding schemes encompassing the evaluation of replicated clones and half‐sib (HS), first selfing generation (S1) or second selfing generation (S2) families without or with within‐family selection. The comparison was based on actual gains for dry‐matter yield over 43 mo and final plant survival across irrigated and rainfed cropping of advanced generation synthetics (Syn‐3) built from 12 selected parents. The study envisaged biomass yield selection over 31 mo under dense planting among 125 S0 plants and/or their progenies, after a stage of stratified mass selection which acted as a benchmark for genetic gains and whose genetic progress over the original genetic base was concurrently assessed. Half‐sib progeny‐based selection exhibited highest genetic gains with respect to unit time and return on yearly investment for direct costs for both biomass yield and crop persistence. Among‐and‐within‐HS family selection ranked second, and S1 progeny‐based selection ranked third, for yield gains, whereas three schemes implying two selfing generations were valuable for improving persistence but not yield. The initial stage of stratified mass selection was time‐ and cost‐efficient for improving yield but not persistence. We provided information also on self‐fertility and inbreeding depression variation across selfing generations, genetic variation of clonal, HS, S1 and S2 material, and consistency of S0 parent value across different types of material. Our results may have implications for field‐based and genomic selection.