1. The effects of various sulphonylureas and diazoxide on insulin secretion and the activity of various channels have been studied using tissue culture and patch-clamp methods in an insulin-secreting cell line derived from a rat islet cell tumour. 2. Tolbutamide, glibenclamide and HB699 increased the rate of insulin release by 2-5 fold. The concentrations of tolbutamide and glibenclamide giving half-maximum effects on insulin secretion were approximately 40 microM and 0.2 microM, respectively. 3. Diazoxide (0.6-1.0 mM) per se, had either no effect or produced a small increase in insulin secretion, whereas when secretion was maximally stimulated by the combination of glucose (3 mM) and leucine (20 mM), it produced inhibition. Tolbutamide-induced release was also inhibited by diazoxide. 4. Tolbutamide, glibenclamide, HB699 and HB985 reduced the open-state probability of the ATP-K+ channel in a dose-dependent manner. Tolbutamide and glibenclamide were shown to be effective regardless of which side of the membrane they were applied. 5. In whole cell recording, in which the total ATP-sensitive K+ conductance of the cell could be measured, dose-inhibition curves for tolbutamide and glibenclamide were constructed, resulting in Ki values of 17 microM and 27 nM, respectively. The value of Ki for tolbutamide was unchanged when ATP (0.1 mM) was present in the electrode. 6. Diazoxide (0.6 mM) activated the ATP-K+ channels only when they had first been inhibited by intracellular ATP (0.1 mM) or bath applied tolbutamide (3-30 microM). The inhibition produced by glibenclamide could not be reversed by diazoxide. 7. Neither tolbutamide (1.0 mM) nor glibenclamide (10 microM) altered the open-state probability of the Ca2+-activated K+ channel or the Ca2+-activated non-selective cation channel which are present in this cell line. 8. It is concluded that the sulphonylureas and related hypoglycaemic drugs and diazoxide regulate insulin secretion by direct effects on the ATP-K+ channel or a protein closely associated with this channel.