This study analyses the convergence of CO 2 emissions at state-level in the USA for the period from 1976 to 2014 in a nonlinear and novel empirical framework. In so doing, we have applied Pesaran's (2007) test of pair-wise approach to testing convergence which gives in general what are the rejection frequencies and thus provides evidence of convergence. At the aggregate level, we also applied Chi-Young et al. (2006) half-life convergence test and the KPSS test with Fourier transformation which states are converging towards a cross-section average. Finally, we also adopted club convergence approach developed by Phillips and Sul (2007) to identify if the states are converging towards a club and last but not least, we applied Schnurbus et al., 2017 test to find if there is possible evidence if club merging. We make two contributions to the literature: (i) we conduct a country-specific analysis by focusing on the US; (ii) we consider both convergence and club convergence. Our overall results from the Pesaran's (2007) pair-wise approach of convergence indicates that about 35% of the time the null of a unit root is rejected when ADF test is used and about 22% of the time null is rejected when ADF-GLS is used (irrespective of AIC or SBIC criterion). These results are also supported by KPSS stationary test which shows that null is rejected about 70–80% times. However, when Fourier function is incorporated in the KPSS test we find that the null hypothesis of stationarity is rejected only for Florida, Massachusetts, Montana, Nevada, New Mexico, Rhode Island, Texas indicating that only these states are non-convergent. Our overall results from club convergence (after club merging) show that USA states are forming 4 clubs. Our findings provide new insight into the convergence of CO 2 emissions at the state level in the USA and thus have profound implications in terms of environmental policy setting and Per Capita Emission (PCE) allocations.