Some wheat breeders conduct few crosses with a restricted number of parents, while others stratify their resources on the progeny from many different crosses among a diverse set of parents. There is admittedly no consensus which strategy is preferable and with a limited budget in breeding programs the progeny populations of specific crosses can be large with the former strategy, while with the latter the number of progenies derived from each cross has to be reduced. A key question is thus how a breeder can influence the genetic gain and variance in this framework by his or her parental selection and in the planning of specific crosses. The aim of this study was accordingly to 1) investigate the importance of parental selection in comparison to the choice of specific crossing combinations, and 2) the influence of different resource allocations on the genetic gain and variance with respect to the number and size of progeny populations. A simulation study was for this purpose conduced based on a founder population of 901 genotyped breeding lines from an applied breeding program. Different subsets of breeding lines and markers were repeatability sampled from this population with the latter representing causal loci to simulate traits with heritabilities of 0.20, 0.60, and 1.00. Simulated progeny populations that varied in their number and size were employed to assess the genetic gain and variance relative to the parental population. The progeny populations themselves were derived either by selecting parents or crosses directly among all possible combinations as well as by random sampling. Similar genetic gain could be achieved by pre-selection the best performing parents irrespective of the resource allocation. Selecting the best crosses among a set of randomly chosen parents was though dependent on the resource allocation as fewer crosses resulted in an higher genetic gain. Pre-selecting parents and the most promising crosses among them gave generally the highest genetic gain, in which a strategy with few large crosses showed the best performance. However, the latter resource allocation also suffered the largest loss of genetic variance. This trade-off between genetic gain and variance leveled off when conducting a larger number of crosses i.e. the short and long-term selection gain can be influenced by a breeders´ resource allocation when planning crosses. Diversifying the risk of failure and loss of variation by conducting many crosses, while testing more progeny of the most promising crosses might accordingly be a convenient strategy.