Using balanced panel data from 188 Chinese cities covering 2006–2017, this study investigated disparities in energy efficiency by considering production technology based on stochastic meta-frontier analysis. A panel quantile regression was conducted to examine and discriminate the heterogeneity of dominant factors in high energy efficiency cities compared to low energy efficiency cities. The results showed that energy efficiency was expected to improve by 24 % when referencing the optimal production technology, but by 16 % when considering discrepancies in technological capability across cities. The factors impacting energy efficiency were significantly heterogenous across quantiles. Specifically, the effects of economic development and foreign capital utilization on energy efficiency were significantly stronger in the lower quantile cities compared to the upper quantile cities. Urbanization had a significant impact on the meta-frontier energy efficiency for all quantile cities, but was insignificant with respect to group-frontier energy efficiency. Industrial structure was the largest factor inhibiting both group-frontier and meta-frontier energy efficiency in all quantile cities. Thus, optimizing industrial structure is an effective way to improve energy efficiency at the city level. The study concludes that energy policies should consider disparities in energy efficiency, particularly the heterogeneous effect of drivers in different cities.