Concrete carbonization is one of the major reasons for the degradation of reinforced concrete (RC) structures, which reduces their durability, serviceability and safety. Therefore, the reasonable prediction model of concrete carbonation depth can provide a reference for durability evaluation and service life prediction of RC structures and be of great significance. Firstly, 273 sets of measurements of China natural carbonation depth of concrete from publications in China and foreign journals were collected and depicted with the histogram. According to the Kolmogorov-Smirnov goodness of fit test, it was found that the measurements of carbonization depth follows Gamma probability distribution. Subsequently, in terms of the Chinese national standard GB-T51355-2019 model and the international standard CEB-FIP-2010 fib-model along with the natural environment of RC structure concrete carbonization, the main factors affecting the process of concrete carbonization were analyzed, concluding that concrete aging compressive strength, relative humidity, temperature, curing mechanism, CO₂ concentration, shelter and rain are the main influencing factors. Secondly, due to the fact that the natural carbonization process of RC concrete is a random, monotonic and increasing one, the Gamma stochastic process model was thus used to characterize it. Using the model and the above main influencing factors as explanatory variables, a new Gamma generalized linear regression model for the natural carbonization depth prediction of concrete was established. Finally, the natural carbonation depth of real RC structure concrete was calculated by the proposed model, and compared with the measured data as well as the predictions by the Chinese standard model and the international fib-model. It was shown that the proposed model has better accuracy in predicting carbonation depth of RC structural concrete than the existing national and international standard models do, and can be used as a potential surrogate model.