Background With increases in cannabis use and potency, there is a need to improve our understanding of the impact of use on cognitive function. Previous research indicates long-term cannabis use may have a negative effect on executive function. Few studies have examined persistence of it in protracted abstinence, and there is limited evidence of predictors of worse cognitive function in current and former users. In this study, we aim to evaluate the associations between cannabis use status (current, former, and never use) and self-report cognition. Further, we investigate if cannabis use characteristics predict self-report cognitive function. Methods Cross-sectional cannabis use data from the National Epidemiological Survey on Alcohol and Related Conditions-III (NESARC-III), a national survey (N = 36,309) conducted in the USA between 2012 and 2013 were used alongside the Executive Function Index scales. The data were analyzed by using Ordinary Least Squares regression. Results Current (N = 3,681, Female = 37.7%) and former users (N = 7,448, Female = 45.4%) reported poorer cognition than never users (N = 24,956, Female = 56.6%). Self-reported cognition of former users was in-between that of current and never users. Several cannabis use characteristics were associated with self-reported cognition in current and former users. Conclusion While prospective studies are required to confirm, findings suggest cannabis use is linked to worse cognition. There may be some limited recovery of cognition in former users and some cannabis use characteristics predict impairment. These findings add to our understanding of the cognitive impact of cannabis use. As worse cognitive function may impact relapse, findings have implications for personalization of cannabis use disorder treatment.