Principal components analysis (PCA) and partial least squares regression (PLS) were applied to multivariate descriptions of insulin content and secretion in allografts of endocrine pancreas. Islets from non-inbred ob/ob-mice were transplanted to the kidney of normal NMRI mice. Recipients were treated daily with cyclosporin-A (CsA; 15-25 mg/kg body weight), CsA in combination with verapamil (0.4 mg/kg body weight), or the vehicle for CsA. After 3 weeks the allografts were removed and analyzed in vitro. In the complete data matrix (23 objects and 26 variables), the first two principal components (PC) explained 90% of the variance and separated fresh islets, vehicle-treated grafts and CsA-treated grafts from each other. Variables describing insulin content and the dynamics of glucose-stimulated secretion dominated the first PC, whereas basal release variables carried the heaviest loadings in the second. As compared with fresh islets, vehicle-treated grafts lost 98% of their insulin and did not respond to glucose. The decline in insulin content and release was less pronounced in grafts protected by CsA. Combining CsA with verapamil resulted in further improvements of insulin content and secretion. PLS disclosed relationships between insulin content and release that are indicative of enhanced graft survival in response to CsA and verapamil. Surviving /3-cells exhibited a diminished proportion of first-phase stimulated release, perhaps reflecting the dislocation and denervation of the grafts. It is concluded that verapamil exerts a partially protecting effect on the function of islet allografts and that the usefulness of PCA and PLS should be further explored in the field of experimental diabetes research.