This paper presents evidence of the disruption of a transition from fossil fuels to nuclear power, and finds the benefits forgone as a consequence are substantial. Learning rates are presented for nuclear power in seven countries, comprising 58% of all power reactors ever built globally. Learning rates and deployment rates changed in the late- 1960s and 1970s from rapidly falling costs and accelerating deployment to rapidly rising costs and stalled deployment. Historical nuclear global capacity, electricity generation and overnight construction costs are compared with the counterfactual that predisruption learning and deployment rates had continued to 2015. Had the early rates continued, nuclear power could now be around 10% of its current cost. The additional nuclear power could have substituted for 69,000–186,000 TWh of coal and gas generation, thereby avoiding up to 9.5 million deaths and 174 Gt CO2 emissions. In 2015 alone, nuclear power could have replaced up to 100% of coal-generated and 76% of gas-generated electricity, thereby avoiding up to 540,000 deaths and 11 Gt CO2. Rapid progress was achieved in the past and could be again, with appropriate policies. Research is needed to identify impediments to progress, and policy is needed to remove them.