<p>Genetic variants that increase the risk for complex diseases persist in human populations, despite adverse effects on health and longevity. Life-history theory predicts that such alleles can be maintained by trade-offs arising from pleiotropy, yet direct genomic evidence has been limited. We asked whether disease-associated variants persist because they enhance reproduction, despite costs to health and lifespan. By analysing genome-wide data across 62 diseases, longevity and fertility, we show that disease-risk alleles are, on average, associated with reduced longevity and increased fertility. Moreover, the subset of alleles that increase both fertility and disease risk appear to have been favoured by natural selection over the past 50,000 years. Using Mendelian randomization, we detect a causal effect of genetic liability to disease on longevity, but no robust evidence for a causal effect on fertility; importantly, these estimates remain stable after adjusting for socioeconomic factors. At the individual level, we compared offspring numbers between affected and unaffected individuals with high polygenic disease risk. For most diseases, affected individuals had more children than unaffected ones. But for early-onset diseases, the pattern reverses, indicating reproductive costs of early morbidity. Together, these results support antagonistic pleiotropy and help explain the persistence of disease-risk alleles in human populations.</p>