OBJECTIVE: To estimate a normal reference range for thyroid-stimulating hormone (TSH) at each point in gestation in singleton and twin pregnancies. METHODS: All women enrolling for prenatal care from December 2000 through November 2001 underwent prospective TSH screening at their first visit. Separate nomograms were constructed for singleton and twin pregnancies using regression analysis. Values were converted to multiples of the median (MoM) for singleton pregnancies at each week of gestation. RESULTS: Thyroid-stimulating hormone was evaluated in 13,599 singleton and 132 twin pregnancies. Thyroid-stimulating hormone decreased significantly during the first trimester, and the decrease was greater in twins (both P < .001). Had a nonpregnant reference (0.4-4.0 mU/L) been used rather than our nomogram, 28% of 342 singletons with TSH greater than 2 standard deviations above the mean would not have been identified. For singleton first-trimester pregnancies, the approximate upper limit of normal TSH was 4.0 MoM, and for twins, 3.5 MoM. Thereafter, the approximate upper limit was 2.5 MoM for singleton and twin pregnancies. CONCLUSION: If thyroid testing is performed during pregnancy, nomograms that adjust for fetal number and gestational age may greatly improve disease detection. Values expressed as multiples of the median may facilitate comparisons across different laboratories and populations.