The kinetics of absorption of intraperitoneally administered insulin were studied in nine uraemic insulin-dependent diabetics undergoing continuous ambulatory peritoneal dialysis (CAPD). In each of three studies 20 U of regular insulin was directly injected as a bolus into the peritoneal cavity through an indwelling Tenckhoff catheter. In two procedures the insulin injection was followed by the instillation of either 2 litres of 1.5% dextrose dialysates or 2 litres of 4.5% dextrose dialysate. In the third 20 ml of saline was used to flush the tubing. Plasma free insulin values rose more rapidly and reached significantly higher concentrations (55.6 +/- 18.8 mU/l) when the insulin had been injected into an empty peritoneal cavity than when it was followed by dialysate. These differences were observed despite the fact that most of the insulin injected was retained by the patients. Since the plasma insulin values did not differ after instillations of dialysate containing 1.5% and 4.5% dextrose, the osmolality of the dialysate seemed not to affect insulin absorption, and the dilution of the insulin probably delayed its transfer through the peritoneum. These findings suggest that insulin given intraperitoneally to patients undergoing CAPD will be most effective if it is given into an empty peritoneal cavity at least 30 minutes before the dialysate is instilled.