Prenatal arsenic exposure has been associated with reduced thymic index and increased morbidity in infants, indicating arsenic-related impaired immune function. We aimed at elucidating potential effects of pre- and postnatal arsenic exposure on cell-mediated immune function in pre-school aged children. Children born in a prospective mother-child cohort in rural Bangladesh were followed up at 4.5 years of age (n = 577). Arsenic exposure was assessed by concentrations of arsenic metabolites (U-As) in child urine and maternal urine during pregnancy, using high-performance liquid chromatography online with inductively coupled plasma mass spectrometry. For assessment of delayed type hypersensitivity response, an intradermal injection of purified protein derivative (PPD) was given to Bacillus Calmette-Guerin vaccinated children. The diameter (mm) of induration was measured after 48-72 h. Plasma concentrations of 27 cytokines were analyzed by a multiplex cytokine assay. Children's concurrent, but not prenatal, arsenic exposure was associated with a weaker response to the injected PPD. The risk ratio (RR) of not responding to PPD (induration <5 mm) was 1.37 (95% confidence interval (CI): 1.07, 1.74) in children in the highest quartile of U-As (range 126-1228 μg/l), compared with the lowest (range 12-34 μg/l). The p for trend across the quartiles was 0.003. The association was stronger in undernourished children. Children's U-As in tertiles was inversely associated with two out of 27 cytokines only, i.e., IL-2 and TNF-α, both Th1 cytokines (in the highest tertile, regression coefficients (95% CI): -1.57 (-2.56, -0.57) and -4.53 (-8.62, -0.42), respectively), but not with Th2 cytokines. These associations were particularly strong in children with recent infections. In conclusion, elevated childhood arsenic exposure appeared to reduce cell-mediated immunity, possibly linked to reduced concentrations of Th1 cytokines.