This paper provides new empirical evidence on Environmental Kuznets Curves (EKC) for CO2 and air pollutants at sector level. A panel dataset based on the Italian NAMEA (National Accounting Matrix including Environmental Accounts) over 1990-2006 is analysed, focusing on both emissions efficiency (EKC model) and total emissions (IPAT model). Results show that, looking at sector evidence, both decoupling and also eventually re-coupling trends could emerge along the path of economic development. The overall performance on here CO2, is not compliant with Kyoto targets. SOx and NOx show decreasing patterns, though the shape is affected by some outlier sectors with regard to joint emission-productivity dynamics. Services tend to present stronger delinking patterns across emissions than manufacturing. Trade expansion validates the pollution haven in some cases, but also show negative signs when only EU15 trade is considered: this may due to technology spillovers and a positive 'race to the top' rather than the bottom among EU15 trade partners. General R&D expenditure show weak correlation with emissions efficiency. EKC and IPAT derived models provide similar conclusions overall. Finally, we used SUR estimators (Seemingly Unrelated Regressions) for EKC models on manufacturing to have more efficient panel estimates (constrained model) and to test for slope heterogeneity (unconstrained model): the empirical evidence for CO2 and SOx emissions suggests that of manufacturing the slope varies across sectors. Further research should be directed towards deeper investigation of trade relationship at sector level and increased research into and efforts to produce specific sectoral data on 'environmental innovations'.