CONTEXT: Evidence has shown maternal androgen levels in both the general population and populations with hyperandrogenic disorders are inversely associated with offspring birth weight. CONTEXT: We aimed to investigate the causal effect of maternal testosterone levels in the general population on offspring birth weight and preterm delivery risk using a two-sample Mendelian randomization (MR) method. METHODS: We obtained independent genetic instruments from a sex-specific genome-wide association study with up to 230 454 females of European descent from the UK Biobank. Genetic instruments with consistent testosterone effects but no aggregate effect on sex hormone-binding globulin were used to perform the main analysis. Summary-level data of offspring birth weight adjusted for genotype were obtained from a study with 210 406 females of European descent. Summary-level data of preterm delivery were obtained from the FinnGen study (6736 cases and 116 219 controls). RESULTS: MR analysis showed that each SD (0.62 nmol/L) increase in testosterone levels could reduce the offspring birth weight by 37.26 g (95% CI, 19.59-54.94 g; P = 3.62 × 10-5). Each SD increase in testosterone levels was also associated with an increased risk of preterm delivery (odds ratio = 1.329; 95% CI, 1.161-1.520; P = 3.57 × 10-5). Similar results were found using different MR methods and multivariable MR analyses. CONCLUSION: This two-sample MR study showed genetically determined higher circulating testosterone levels in females from the general population were associated with low birth weight of offspring and increased risk of preterm delivery.