Background: Jaundice is common among neonates and if untreated can lead to kernicterus. Diagnosing of jaundice in neonates using Kramer’s method (visual assessment) is considered user friendly in resource limited areas. However, there are conflicting finding on reliability of the Kramer’s method in diagnosis of neonatal jaundice (NJ) particularly of black descent. Therefore, this study aimed to determine diagnostic accuracy of Kramer’s method in comparison with total serum bilirubin (TSB) test in diagnosis of NJ among neonates of black descent in Tanzania. Methods: A cross-sectional study was conducted between June and July 2020 at Muhimbili National Hospital (MNH) in Dar es Salaam Tanzania. A total of 315 neonates were recruited consecutively. In each neonates’ jaundice was assessed by using Kramer’s method and TSB test. A 2 X 2 table was created for determination of sensitivity, specificity, positive predictive values (PPV), negative predictive value (NPV), positive and negative likelihood ratios (+LR/-LR) and diagnostic accuracy (effectiveness). Cohen kappa (κ) was used to analyze the agreement between Kramer’s method and TSB. Association between independent variables and presence of jaundice were assessed using chi-square test and the p ˂ 0.05 was considered to be statistical significance. Results: The prevalence of NJ was 49.8% by Kramer’s method and 63.5% by TSB. The Sensitivity, Specificity, PPV, and NPV of the Kramer’s method were 70.5%, 86.1%, 88.8%, and 62.6%, respectively. The +LR and –LR were 5.07 and 0.34 respectively. The diagnostic accuracy of the Kramer’s method was 76.1%. There was a moderate agreement between Kramer’s method and TSB results (κ= 0.524, P˂0.001). No significance relationship between the independent variables and presence of NJ. Conclusion: Kramer’s method was found to be inefficient in detecting NJ among neonates of black descent. However, it can be used as a predictor of NJ and whenever available invasive techniques should be applied.