Iodine is a vital trace element in the human body and is associated with several important coronary artery disease (CAD) risk factors. We aimed to explore the correlation between urinary iodine concentration (UIC) and CAD. Data from 15 793 US adults in the National Health and Nutrition Examination Survey (2003–2018) were analysed. We conducted multivariable logistic regression models and fitted smoothing curves to study the correlation between UIC and CAD. Furthermore, we performed subgroup analysis to investigate possible effect modifiers between them. We found a J-shaped association between UIC and CAD, with an inflection point at Lg UIC = 2·65 μg/l. This result indicated a neutral association (OR 0·89; 95 % CI 0·68, 1·16) between UIC and CAD as Lg UIC < 2·65 μg/l, but the per natural Lg [UIC] increment was OR 2·29; 95 % CI 1·53, 3·43 as Lg UIC ≥ 2·65 μg/l. An interaction between diabetes and UIC might exist. The increase in UIC results in an increase in CAD prevalence (OR 1·84, 95 % CI 1·32, 2·58) in diabetes but results in little to no difference in non-diabetes (OR 0·98, 95 % CI 0·77, 1·25). The J-shaped correlation between UIC and CAD and the interaction between diabetes and UIC should be confirmed in a prospective study with a series of UIC measurements. If excessive iodine precedes CAD, then this new finding could guide clinical practice and prevent iodine deficiency from being overcorrected.