Reliable emission estimate of non-methane volatile organic compounds (NMVOCs) is important for understanding the atmospheric chemistry and formulating control policy of ozone (O3). In this study, a speciated emission inventory of anthropogenic NMVOCs was developed with the refined “bottom-up” methodology and best available information of individual sources for Nanjing in 2017. The total NMVOCs emissions were calculated at 163.2 Gg. It was broken down into the emissions of over 500 individual species and aromatics took the largest fraction (33.3% of the total emissions). Meanwhile, 105 compounds were measured at 5 sites representing different functional zones of Nanjing for one year. The annual mean concentration of totally 105 species varied from 48.5 ppbv to 63.7 ppbv, and alkanes was the most abundant group with its mass fractions ranging 37.2–40.1% at different sites. Constrained by the emission ratios of individual species versus carbon monoxide (CO) based on ground measurement, the total emissions of 105 species (NMVOCs-105) was estimated at 195.6 Gg, 81.1% larger than the bottom-up estimate of NMVOCs-105 (108.0 Gg). The constrained emissions indicated an overestimation of aromatics and underestimation of OVOCs and halocarbons in the bottom-up emission inventory because of the uncertainties in source profiles. O3 simulation with Community Multi-scale Air Quality (CMAQ) model was conducted for January, April, July and October in 2017 to evaluate the bottom-up and constrained emission estimates. The mean normal bias (MNB) and mean normal error (MNE) values were generally within the criteria (MNB ≤ ±15% and MNE ≤ 30%) for both inventories. The model performance was improved when the constrained estimates were applied, indicating the benefit of ground observation constraints on NMVOCs emission estimation and O3 simulation. Based on the O3 formation potential (OFP), 12 key NMVOCs species mainly from surface coating, on-road vehicles and oil exploitation and refinery were identified as the priority compounds for O3 reduction.