We studied the variation in plasma insulin-like growth factor-I (IGF-I) concentrations in unrestrained, cannulated rats. To detect rapid changes, 6 rats were sampled every 15 min for 6 h. Plasma was assayed for growth hormone (GH), and for IGF-I before and after acid-ethanol (AE) extraction to reduce the masking effect of plasma binding proteins. AE increased the immunoreactivity of freshly collected rat serum and heparinized plasma by 3-fold, but had no effect on serum after 4 weeks of storage at -20 degrees C. In contrast, heparinized plasma maintained its sensitivity to AE for many months after collection. AE-extracted serum was identical in potency to serum that was subjected to acid gel chromatography. GH showed high amplitude, synchronized pulses every 3.3 +/- 0.15 h. Despite variations in IGF-I concentrations that resembled pulsations, when the data from both unextracted and AE-extracted plasma were subjected to 2 computer algorithms designed to detect hormone pulses, no pulses were identified in 8 of the 24 series of analyses. The two programs concurred only 4 times in their identification of a pulse in unextracted plasma, and in only one instance was a pulse identified in both unextracted and AE-extracted plasma from a given sample. Based on repeated measurements of a single serum sample, the program of Santen and Bardin (S&B) had a false-positive rate of 11%, and that of Merriam and Wachter (M&W) less than 10%. There was no positive correlation between summed GH and IGF-I concentrations over the 6 h of sampling, or after a 1- or 2-hour lag period. To detect diurnal variations, a second group of 5 rats was sampled every 2 h for 36 h. These animals showed little fluctuation of IGF-I concentrations and no differences between the light and dark periods. Our studies provide no evidence for episodic release of IGF-I or diurnal variations entrained to light or feeding cycles.