Because of difficulties encountered in setting up radioimmunoassays for cholecystokinin (CCK) a sensitive and reliable biological method for estimating this hormone is still needed. The principles of such a biological technique and an improvement to it have already been described, but the serum levels of CCK reported were high and the technique required further refinement and validation. The strips of rabbit gall-bladder used to estimate the concentration of CCK increased in sensitivity to standard solutions of CCK over a 6--8 period before stabilizing, but a single sample of serum increased the sensitivity of the strips of gall-bladder to their maximum immediately. These two problems were eliminated by 'priming' the strips of gall-bladder by exposure to two serum samples before exposure to the standard solutions used for production of a dose--response curve. Thirdly, it was discovered that some non-peptide substances in serum possessed CCK-like activity; by extracting all the small peptides from serum with dextran-coated charcoal the residual activity could be measured and subtracted from the total CCK activity. Finally, the activity of CCK in the serum increased during processing before freezing. This increase was eliminated by taking the blood samples into aprotinin which has been shown to cause dramatic reduction in CCK activity in some experiments. When all these factors were taken into account and the technique suitably modified, the mean level of CCK in the serum of ten normal fasting subjects was found to be 28 milli Ivy Dog units/ml (2.4 pmol/ml), which is only one third of that reported previously.