Background: The creatinine-based CKD EPI equation is the most widely used method to estimate glomerular filtration rate (eGFRcr) in clinical practice. The eGFRcr adjusts for sex, age, and race (Black vs non-Black) and was derived and validated in aggregated datasets from US-based research studies in which GFR was measured directly. However, in prior work, we found eGFRcr performed worse in people living with HIV (PLWH) compared to HIV negative counterparts which may be due to differences in muscle mass, the major source of creatinine. Here, we focus on African American (AA) participants to determine whether the race eGFRcr calibration factor contributes to poor accuracy and bias in AAs living with HIV. Methods: Beginning in 2010, we enrolled and followed an AA cohort of PLWH and HIV-negative participants in Baltimore Maryland. Annually, we measured GFR by iohexol disappearance from plasma (iGFR) and serum concentrations of creatinine and cystatin C. This analysis was restricted to AA participants. We calculated eGFRcr and the creatinine-cystatin C combination equation (eGFRcr-cys) with and without race adjustment. Accuracy was defined as the fraction of eGFR values within +/- 30% of iGFR. Bias was defined as eGFR minus iGFR, with positive and negative values representing overestimation and underestimation of iGFR, respectively. We used multilevel mixed models to account for the within-visit linked structure of the multiple GFR measures, further nested within repeated observations for individuals. We examined the association between lean mass, HIV status, and eGFRcr bias in a subset with body composition measures. Findings: 207 HIV-positive and 107 HIV-negative AA participants contributed 781 and 376 study visits, respectively, with valid measures of iGFR, creatinine, and cystatin C. Among PLWH, omitting the race adjustment (compared with retaining it) changed average eGFRcr bias from 9.1 to -3.9 ml/min/1.73 m2. Moreover, estimation accuracy improved significantly when race adjustment was omitted rather than retained: 86% vs. 78% for eGFRcr (P<0.001) and 91% vs. 88% for eGFRcr-cys (P=0.045). Lean mass (but not other measures of body size or composition) was correlated with eGFRcr bias and, in adjusted analyses, lean mass was significantly lower in PLWH compared with HIV-negative AAs compatible with not using the race coefficient. Interpretation: We found that the standard, widely used eGFRcr equation as well as eGFRcr-cys overestimate iGFR and has poor accuracy in AAs living with HIV. Omitting the race coefficient improved both bias and accuracy. HIV clinicians should be aware that race adjustment may worsen the performance of creatinine based GFR estimates in AAs. Funding Statement: This study was supported by the National Institute on Drug Abuse of the National Institutes of Health (R01 DA026770, K24 DA035684) and by the Johns Hopkins University Center for AIDS Research (P30 AI094189). MGA was supported by National Institute of Diabetes and Digestive and Kidney Diseases (P01 DK056492). GE Healthcare provided iohexol to the study free of charge. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. Declaration of Interests: None to declare. Ethics Approval Statement: The study was approved by the Johns Hopkins University School of Medicine Institutional Review Board, and participants provided written informed consent.