1 The MIN6 cell line derived from in vivo immortalized insulin-secreting pancreatic beta cells was used to study the insulin-releasing capacity and the cellular mode of action of S-21663, a newly synthesized imadizoline compound known for its antidiabetic effect in vivo and its ability to release insulin from perfused pancreas. 2 S-21663, at concentrations ranging from 10(-5) M to 10(-3) M was able to release insulin from MIN6 cells; its activity peaked at 10(-4) M, a drop in the stimulant factor being noted between 10(-4) and 10(-3) M. Its efficacy, which did not differ whatever the glucose concentration (stimulant or not), was higher than that of the other secretagogues tested, glucose, sulphonylureas or the peptide tGLP-1. 3 In contrast to tGLP-1, S-21663 did not change the cyclic AMP content, whereas it increased Ca2+ influx via verapamil-and nifedipine-sensitive voltage-dependent calcium channels, the insulin release being a direct consequence of this Ca2+ entry. The S-21663-induced Ca2+ influx appears to be essentially the consequence of closure of K+ channels which differ from the ATP-dependent K+ (K-ATP) channels as determined by measurement of Rb-86 efflux and use of a K-ATP channel opener. 4 Comparison of the effects of S-21663 to that of efaroxan, another imidazoline compound shown to act on insulin release in a glucose-dependent way via binding sites distinct from the imidazoline I-1 and I-2 sites. suggested that S-21663 acts through a novel site which displays a remarkably stable expression along the cell culture. 5 It is concluded that S-21663 is a very efficient, glucose-independent insulin secretagogue acting through a novel imidazoline site, linked to K+ channels, distinct from the I-1, I-2 and 'efaroxan' binding sites. In vitro and in vivo features of S-21663 indicate that this compound, or new drugs drived from it, might be the basis for a new pharmacological approach to the mangement of type II (non insulin-dependent) diabetes.