Our study contributes to the empirical literature in two ways: 1) We propose an innovative Data Envelopment Analysis (DEA) model to estimate the cost efficiency of Chinese banks; 2) We propose a Stochastic Structural Relationship Programming (SSRP) Model based on neural networks to evaluate the interrelationships between efficiency and other bank-specific variables. The findings show that Chinese commercial banks experienced a gradual increase in the level of efficiency from 2010 up to 2015, after which the efficiency level experienced slight volatility and, by the end of 2018, ended up with an efficiency score of 0.746 on a scale of 0 to 1. We find that for banks with lower levels of efficiency, improvement in the efficiency level has a favorable effect on profitability, and these banks will also focus their business on traditional banking activities. Banks with higher levels of efficiency need to find alternative, profitable banking businesses in order to sustain their higher level of efficiency.