Abstract Crossing over breaks linkages and leads to a wider array of allele combinations. My objective was to assess the contribution of crossing over to genetic variance ( V G ) in maize ( Zea mays L.) and wheat ( Triticum aestivum L.). The contribution of crossing over to V G (denoted by P CO ) was assessed by calculating V G without crossing over from the sums of marker effects on each chromosome and by estimating V G with crossing over from simulated doubled haploids that arise from meiosis. For maize yield, crossing over had positive contributions of P CO = 7% and 16% in two populations but it strongly decreased V G ( P CO = −74% to −25%) in five other populations. The mean P CO was negative for moisture, test weight, plant height, and ear height. In wheat, the P CO values were all negative for five traits in the Louise/Penawawa population but were all positive for three traits in the Seri/Babax population. Negative P CO values were attributed to large differences between the sum of allelic effects on a homolog inherited from one parent and the sum of effects on the homolog from the other parent. Although crossing over most often decreased V G , the best simulated line (out of 10,000) with crossing over was usually superior to the best line without crossing over. Breeding progress will therefore continue to rely on finding individuals with increasingly rare, favorable crossovers amidst individuals with crossovers that are mostly unfavorable.