OBJECTIVES To evaluate the effectiveness of thyroid-stimulating hormone (TSH) and thyroxine (T4) measurements at neonatal screening for congenital hypothyroidism, we compared our false-negative results with those we would have obtained if we had used TSH screening alone. SUBJECTS AND METHODS Between January 1989 and December 2001 745,258 newborns were screened (98.3% of total born) for congenital hypothyroidism in northeast Italy. T4 and TSH were measured simultaneously on blood spots collected after birth. Between 1989 and 1998, semi-quantitative total T4 (tT4) and TSH concentrations were measured by radiolabelled immunological assay and, from 1999 to 2001, using time-resolved fluorometer Delfia instruments (EG&G Wallac Oy, Finland) and fluoroimmunometric assay (Delfia neonatal hTSH and T4 kits). RESULTS Ten neonates were missed by our screening programme (normal tT4 and TSH) and classified as false negatives; these infants were diagnosed later in life with central hypothyroidism. If we had measured TSH alone in our screening programme, we would have missed an additional 21 patients with low tT4 and normal TSH; of these, four were affected by central hypothyroidism and 17 were diagnosed within the second month of life as affected by primary hypothyroidism with delayed TSH rise. CONCLUSIONS Simultaneous T4 and TSH measurements at neonatal screening can miss patients affected by central hypothyroidism. However, this screening procedure allows identification of cases of central hypothyroidism with low T4 values and those neonates affected by primary hypothyroidism with delayed TSH rise who we would have missed by using the TSH measure alone.