We used a receiver operating characteristic (ROC) plot to evaluate the diagnostic accuracy of primary thyrotropin (TSH) screening, and compared it to the primary thyroxine (T4) with secondary TSH screening (T4/TSH) on a certain percentage of the lowest T4 value for newborn congenital hypothyroidism (CH). There were 2198 normal and 117 abnormal CH cases evaluated using both the Wallac Delfia neonatal TSH and neonatal T4 kits. The ROC areas of the primary TSH screening and T4/TSH screening were 0.9841 and 0.9557, respectively. Nine cases (out of 117 cases) of CH would have been misclassified if T4/TSH screening were used. These nine cases, however, were identified using primary TSH screening. We conclude that using primary TSH screening is more effective for mass screening in reducing false-positive and false-negative cases than the combined T4/TSH screening method.