This study provides stepwise benchmarking practices of each port to enhance environmental performance using joint application of data mining technique referred as Kohonen's Self-Organizing Map (KSOM) and Recursive Data Envelopment Analysis (RDEA) to address the limitation of conventional DEA. A sample of 20 container ports in the U.S were selected, and data on input variables (number of quay crane, acres, berth and depth), output variables (number of calls, throughput and deadweight tonnage, and CO2 emissions) are used for data analysis. Among the selected samples, eight container ports are found to be environmentally inefficient. However, there appears to be a high potential to become environmentally efficient port. In conclusion, it can be inferred that, stepwise benchmarking process using two combined methodologies substantiates that, more applicable benchmarking target set of Decision Making Units (DMUs) is be projected that consider similarity of physical and operational characteristics of homogenous ports for improving environmental efficiency.