This study investigates 30 provinces in China between 2003 and 2014. Kernel density method is used to analyse the dynamic evolution of haze pollution and technological innovation research and development (R&D), while spatial econometric analysis is used to study the impact of technological innovation on the haze pollution. The results show that haze pollution presents global spatial autocorrelation and local spatial cluster in China. China’s haze pollution has a significant spatial dependence and spatial spillover. A disproportion distribution pattern of haze pollution exists among provinces in China: the central region is the most polluted area followed by the western region, the northeast region and the eastern region. The kernel density curve shows that the gap between technology innovations R&D among provinces has expanded year by year. There is a polarization between the technological innovations R&D. Dynamic evolution results showed that during 2003–2012, the kernel density distribution curve of haze pollution showed a leftward shift, indicating that provincial haze pollution decreased gradually. However, the kernel density distribution curve of haze pollution showed a rightward shift in 2014, and the provincial haze pollution increased. During 2003–2012, the gap of haze pollution among different provinces in China gradually narrowed, while in 2014, the gap increased significantly. Spatial econometric results show that the indirect effects and the total effects of technological innovation are significantly negative. Technological innovation can not only reduce the regional haze pollution but also indirectly lead to the decline in the haze pollution of adjacent provinces through the knowledge spillover effect. The increase of population density can effectively reduce the haze pollution. There is an inverted ‘U type’ relationship between economic development and haze pollution. The increase of traffic pressure will aggravate the degree of haze pollution.