The construction, calibration and evaluation of a prompt- gamma in vivo neutron activation analysis (IVNAA) facility for the simultaneous measurement of total body nitrogen (TBN) and chlorine (TBCl) in children is described. Subjects are irradiated unilaterally by a 0.2 GBq 252Cf neutron source from shoulder to mid thigh in both supine and prone positions. Prompt gamma-ray spectra are measured with two pairs of NaI(TI) crystals (each crystal: 10 cm x 10 cm x 15 cm) positioned on both sides of the subject. TBN and TBCl are estimated from the ratios of nitrogen-to-hydrogen (Nc/Hc) and chlorine-to-hydrogen (Clc/Hc) counts as determined from the measurement of 10.83 MeV, 8.57 MeV and 2.22 MeV prompt gamma-rays from the respective reactions 14N(n, gamma)15N, 36Cl(n, gamma)35Cl, and 1H(n, gamma)D. Nc/Hc and Clc/Hc are corrected for the effect of body width and thickness on background and gamma-ray attenuation. Total body hydrogen (TBH) is used as an internal standard which is independently determined using a four compartment model of body weight defined as the sum of total body water (TBW) measured by the D2O dilution technique, total body protein (TBPr) (i.e. 6.25 x TBN) measured by IVNAA, total body bone mineral (TBBM) measured by dual energy x-ray absorptiometry and total body fat (TBF) estimated as body weight less the sum of TBW, TBPr and TBBM. The effective dose equivalent to a small child is 0.25 mSv (Q = 20) per measurement scan. Repeated measurements of a child-size bottle phantom containing tissue-equivalent concentrations of nitrogen and chlorine yield respective intra- and inter-assay precision values of 2.8% (CV) and 2.3% for TBN measurements, and 7.9% and 10.0% for TBCl measurements. Similarly, intra- and inter-assay accuracy is determined to be respectively +0.1% +/- 1.0% (mean, 95% confidence interval) and +1.4% +/- 1.4% for TBN measurements, and +2.3% +/- 4.3% and +3.9% +/- 6.0% for TBCl measurements.