Class III cardiovascular device premarket approval (PMA) studies often fail to fully represent the intended-use population (IUP) owing to low enrollment of racial and ethnic minority subjects and women. The impact of research site selection on this is unknown. In this study, we sought to determine if site characteristics predict enrollment of demographic minority and female participants in coronary stent PMA trials and evaluate if site selection could improve representation of the IUP. We pooled data from 8,859 U.S. participants enrolled in 9 pivotal coronary stent PMA studies (2003-2018) across 196 sites. Site characteristics included U.S. region, surrounding county demographics, teaching status, Veterans Administration affiliation, trial volume, female principal investigator (PI) involvement, and number of acute hospital beds. Multivariable regression identified predictors of minority and female enrollment. Participant-to-prevalence ratios (PPRs) were modeled under varying site selection scenarios. Minority participants (12%; PPR = 0.48) and women (30%; PPR = 0.77) were underrepresented. Minority enrollment varied markedly across sites and was predicted by West and South regions, county minority population, population density, and per-capita income (R2 = 0.50; P < 0.001). Modeling estimated that reallocating enrollment from low to high minority-enrolling sites could normalize Black and Hispanic representation (PPRs ≥0.80) without compromising that of non-Hispanic Whites (PPR = 1.00). Female enrollment showed less variation and was poorly predicted by research site characteristics and site PI gender (non-VA status only; R2 = 0.095; P < 0.001); however there were few female PIs (<6%), limiting correlation. Coronary stent PMA studies do not fully reflect the IUP, owing to marked underrepresentation of minority participants and modest underrepresentation of women. Because minority enrollment is influenced by site characteristics, targeted site selection could improve representation; however, improving female enrollment requires alternative strategies. These insights have implications on the planning and design of future cardiovascular device trials.