Background: The metabolic syndrome (MetS) is associated with elevated urinary albumin (UA) excretion and C-reactive protein (CRP). However, potential differences in CRP levels on the association between individual components of the MetS and microalbuminuria (MA; 30-300 μg/mL) and/or UA (0-300 μg/mL) by race/ethnicity is unknown. Methods: We analyzed National Health and Nutrition Examination Surveys (NHANES) data, (1999-2010) for adults (≥20 years of age) with the MetS (N = 5700). The Sobel-Goodman mediation test examined the influence of CRP on the association between individual MetS components and both MA and UA by race/ethnicity. We applied machine learning models to predict UA. Results: CRP mediated the association between waist circumference (WC) and MA in Whites and Hispanics but not in Blacks. However, in general, the proportion of the total effect of MetS components on UA, mediated by CRP, was: 11% for high-density lipoprotein cholesterol (HDL-C) and 40% for WC (P < 0.001). In contrast to MA, the mediation effect of CRP for WC and UA was highest for Blacks (94%) compared with Whites (55%) or Hispanics (18%), P < 0.05. The prediction of an elevated UA concentration was increased in Blacks (∼51%) with the MetS when CRP was added to the random forest model. Conclusions: CRP mediates the association between UA and both HDL-C and WC in Whites and Blacks and between UA and WC in Hispanics. Moreover, the machine learning approach suggests that the incorporation of CRP may improve model prediction of UA in Blacks. These findings may favor screening for CRP in persons with the MetS, particularly in Blacks.