Many countries have achieved moderate to dramatic growth during the last few decades, and the world-widely continued economic growth results in increased wealth and deteriorated environment. The relationship between economic growth and environmental quality has received great attention in empirical and theoretical studies. But results are mixed: some find that economic development inevitably leads to environmental deterioration due to resource depletion and pollution, while others find that continued economic development helps to improve environmental quality. Further works are required in a specific context to answer the question whether environmental improvement is compatible with continued economic growth. We intend to provide insight on the potential for carbon emissions control within one nation in the absence of international agreement. The electricity generation sector in China was chosen to demonstrate the economic impact of possible emissions control on the production of power industry in this study. The important economic questions include: What are the optimal policies to provide firms with the incentive to abate carbon emission? Does taxation or subsidy work? What are the impacts of emission control on economic growth? An endogenous growth model is set to represent a regulated power industry , and a Coub-Douglas production function is used to describe the joint production of electricity and carbon emissions. The modified Hamilton approach is employed to solve the model under three possible polices: emission fee, coal tax and abatement subsidy. The theoretical analysis suggests that firms have no incentive to abate in the absence of regulation, and finds that the ratio of emissions to desired output is not a constant, but a function of productive capital and other parameters. The non-constant ratio provides the theoretical grounds to choose the appropriate policies for emissions control. And the sustainable growth could be achieved when appropriate environmental instruments are chosen. Moreover, the optimal conditions derived from the model, rather than ad hoc specification, are used to examine the relationship between desired output and emissions for empirical analysis. Data comes from the China Statistical Yearbook and China Electric Power Yearbook, providing the provincial information for the period 1993-2003. Joint estimation of emissions and output is done using full information maximum likelihood (FIML) method. The optimal emission tax rates are estimated as 59.78%, 13.78% and 6.88% corresponding to abatement efficiencies of 1%, 5% and 10%, respectively. The optimal input tax rates ( ) are 54.05%, 56.13%, 57.55% 65.11% corresponding to discount rate ( ) of 0.05, 0.08, 0.10, 0.20 respectively. The results suggest that the sustainable growth could be achieved when appropriate environmental instruments are chosen, and emission tax is preferred to input tax in the sense of social cost, whenever detection of emissions is not costly and abatement technology allows removing at least 5% of total emissions at pipe end.