Forty-eight children, referred for evaluation of short stature, underwent 24-hour spontaneous growth hormone (GH) secretion studies. The GH level in pooled sera was assessed for each child, using up to 11 commercial immunoassays. In a group of 15 children, the mean GH values obtained by nine of the assays were compared with the mean value given by a polyclonal radioimmunoassay (RIA) from Sorin: four gave higher results (p less than 0.0001), three gave comparable results and two gave lower results (p less than 0.001). The assay yielding the highest results (Nichols: 5.9 +/- 2.3 ng/ml, mean +/- SD) gave values that were approximately triple those obtained by the assay yielding the lowest results (Hybritech: 1.8 +/- 0.8 ng/ml; p less than 0.0001); both of these are monoclonal immunoradiometric assays (IRMAs). The GH concentrations measured in 24-hour pools from 32 children using a monoclonal IRMA from Biomérieux were similar to those obtained using a polyclonal RIA from Farmos (2.8 +/- 1.1 ng/ml and 2.9 +/- 1.4 ng/ml, respectively) but significantly lower than those measured by another polyclonal RIA from Sorin (3.5 +/- 1.5 ng/ml). Two polyclonal assays (Biomérieux and Sorin) were then used to measure the GH levels in all of the 30-minute samples and in the day, night and 24-hour pools from the secretion studies of 22 children. The ratio of the results of the two assays remained fairly constant for a given child (although the GH levels in different 30-minute samples differed considerably). However, the ratios between different children showed quite wide variation (from 2.03 to 1.04).(ABSTRACT TRUNCATED AT 250 WORDS)