This study aims to analyze the nonlinear relationship between environmental regulation and carbon emission efficiency and provide scientific reference for achieving the goal for carbon neutrality at a lower cost. Taking 30 provinces in China, using dual carbon policy as the research objects, the slacks-based measure–Malmquist–Luenberger (SBM–ML) index method was used to measure the carbon emission efficiency from 2009 to 2019 and a panel threshold regression model was established to explore the nonlinear effects of environmental regulation and carbon emission efficiency in each province. The results show that: (1) during the sample period, there is geographical variability in CEE, with the eastern coastal provinces having the highest CEE, followed by the central and western provinces, and the resource-dependent provinces having the lowest CEE and their energy consumption and utilization efficiency being significantly lower than other provinces; (2) when the energy consumption intensity is used as a threshold variable, the relationship between environmental regulation and carbon emission rate is an inverted “U” shape; and (3) when green technology innovation is used as a threshold variable, the relationship between environmental regulation and carbon emission rate is a “U” shape. This study provides a new perspective for improving carbon emission efficiency.