Identifying prenatal environmental factors that have genuinely causal effects on psychopathology is an important research priority, but it is crucial to select an appropriate research design. In this review we explain why and what sorts of designs are preferable and focus on genetically informed/sensitive designs. In the field of developmental psychopathology, causal inferences about prenatal risks have not always been based on evidence generated from appropriate designs. We focus on reported links between maternal smoking during pregnancy and offspring attention-deficit/hyperactivity disorder or conduct problems. Undertaking a systematic review of findings from genetically informed designs and “triangulating” evidence from studies with different patterns of bias, we conclude that at present findings suggest it is unlikely that there is a substantial causal effect of maternal smoking in pregnancy on either attention-deficit/hyperactivity disorder or conduct problems. In contrast, for offspring birth weight (which serves as a positive control) findings strongly support a negative causal effect of maternal smoking in pregnancy. For maternal pregnancy stress, too few studies use genetically sensitive designs to draw firm conclusions, but continuity with postnatal stress seems important. We highlight the importance of moving beyond observational designs, for systematic evaluation of the breadth of available evidence and choosing innovative designs. We conclude that a broader set of prenatal risk factors should be examined, including those relevant in low- and middle-income contexts. Future directions include a greater use of molecular genetically informed designs such as Mendelian randomization to test causal hypotheses about prenatal exposure and offspring outcome.