In order to accurately reveal the spatial difference and time evolution characteristics of total factor productivity of high-tech enterprises in China''s urban agglomerations, this paper is based on the review of existing relevant literature, conducted empirical analysis by using the methods of transcendental logarithmic stochastic frontier production function, Theil index, Kernel density estimation and spatial Markov chain estimation. (1) The results show that the total factor productivity of high-tech enterprises in 19 urban agglomerations shows an overall growth trend, showing obvious regional differences. The overall differences, intra-regional and inter-regional differences all show an expanding trend, and the intra-regional differences are the root cause of the overall differences. (2)The center of the distribution curve of total factor productivity growth shifts to the right, with a trend of trailing to the right. The width of the curve interval is significantly expanded, and the peak value of the main peak decreases significantly, but the polarization phenomenon is different. (3)Considering the spatial correlation effect, the transfer characteristics of TFP of high-tech enterprises in 19 urban agglomerations show obvious spatial correlation, and the transfer probability matrix shows different spatio-temporal evolution characteristics under different proximity levels.The research conclusion has important theoretical and practical significance for reducing the regional difference of total factor productivity growth and promoting the development of regional collaborative innovation.