BACKGROUND: Screening for congenital hypothyroidism (CH) using cord blood or heel-stick samples is considered essential for the prevention of long-term complications CH, which include intellectual disability and slow growth.OBJECTIVE: Compare the sensitivity and specificity of cord blood and heel-stick samples for determining thyroid-stimulating hormone (TSH) levels for the detection of CH.DESIGN: Comparative diagnostic accuracy.SETTINGS: Tertiary care center in Riyadh.PATIENTS AND METHODS: The study included all infants who were delivered during the period from May 2011 to May 2013. As part of routine newborn screening, both cord blood and heel-stick samples were collected from each newborn for CH screening by measuring TSH levels. A cord TSH level was considered positive if the concentration of TSH was more than 60 mIU/L and negative if less than 30 mIU/L. Any cord TSH level between 30-60 mIU/L was considered borderline, and free T4 was measured from the same cord sample. The result was considered positive if the free T4 level was below 9 pmol/L. Heel-stick TSH levels more than 20 µU/L were considered positive. All newborns with positive results were recalled and a peripheral venous sample was taken for TSH and free T4 for confirmation.MAIN OUTCOME MEASURES: Sensitivity and specificity, positive and negative predictive values and recall rates.SAMPLE SIZE: 17 729 screened babies.RESULTS: Of 17 729 neonates screened, 7 were diagnosed as having primary CH. All confirmed cases were detected by both cord and heel-stick TSH levels: 88 cord results were positive (sensitivity 100%, specificity 99.6%, with a recall rate of 0.04%) and 305 heel-stick results were positive (sensitivity 100%, specificity 98.3%, with a recall rate of 1.68%).CONCLUSION: Both cord and heel-stick TSH testing detected all cases of CH. Cord testing was superior to heel-stick testing as the recall rate was lower. We think cord TSH testing is preferable when heel-stick is difficult or early discharge is the practice.LIMITATIONS: Retrospective; the timing of newborn screening for TSH sampling was premature.CONFLICT OF INTEREST: None.