Plant-based diets, particularly when rich in healthy plant foods, have been associated with a lower risk of type 2 diabetes and cardiovascular disease. However, the impact of plant-based diets that distinguish between healthy and unhealthy plant foods on cardiometabolic biomarkers remains unclear. Dietary information was collected by two 24-h recalls among 34,785 adults from a nationwide cross-sectional study. Plasma levels of insulin, C-peptide, glucose, C-reactive protein (CRP), white blood cell (WBC) count, triglycerides (TG), total cholesterol (TC), high-density lipoprotein cholesterol (HDL-C) were measured. Linear regression was used to evaluate the percentage difference in plasma marker concentrations by three plant-based diet indices, namely the overall plant-based diet index (PDI), the healthful PDI (hPDI), and the unhealthful PDI (uPDI). Greater hPDI-adherence scores (comparing extreme quartiles) were associated with lower levels of insulin, Homeostatic Model Assessment of Insulin Resistance (HOMA-IR), TG/HDL-C ratio, CRP, WBC count, and TG, and higher levels of HDL-C, with the percentage differences of - 14.55, - 15.72, - 11.57, - 14.95, - 5.26, - 7.10, and 5.01, respectively (all Ptrend ≤ 0.001). Conversely, uPDI was associated with higher levels of insulin, C-peptide, HOMA-IR, TG/HDL-C ratio, CRP, WBC count, and TG, but lower HDL-C, with the percentage differences of 13.71, 14.00, 14.10, 10.43, 3.32, 8.00, and - 4.98 (all Ptrend ≤ 0.001), respectively. Overall PDI was only associated with lower levels of CRP and WBC count (all Ptrend ≤ 0.001). Our findings suggest that hPDI may have positive, whereas uPDI may have negative impacts on multiple cardiometabolic risk markers, and underscore the need to consider the quality of plant foods in future PDI studies.