The field of reproductive medicine has been criticized for introducing ARTs without systematic research on possible safety risks and for failing to meet the standards of evidence-based innovation held elsewhere in medicine. In this paper, firstly, we ask whether 'responsible innovation' has been a concern for the field, and if so, how it has understood the practical implications of this idea for the development and introduction of potentially risky new ARTs. Secondly, we consider whether the field has indeed fallen short of its responsibilities in this respect, and if so, how things can be improved. To answer these questions, we present three case studies involving the introduction of a new reproductive technology: ICSI, preimplantation genetic testing and mitochondrial replacement therapy. As a framework for analyzing these cases, we used Per Sandin's account of the four dimensions of dealing with risks (threat, uncertainty, action, command) that are central to debates about the possible role of the so-called precautionary principle. We conclude that, although offspring safety concerns have been on the agenda of the debate about bringing the relevant technologies to the clinic, systematic safety and effectiveness studies were not always conducted. As professionals in assisted reproduction have a responsibility to take account of the welfare of the children they are creating, we suggest a policy of proceeding with systematic caution. Legal measures may be needed to ensure that professional guidance is followed in practice. Finally, an open question concerns the threshold for acceptable risk in the context of introducing new ARTs. Multiple stakeholders, including professional societies and patient organizations, should have a role in the urgent debate about this.