The two major forms of infantile persistent hyperinsulinemic hypoglycemia require different treatments, but are difficult to differentiate during surgery. Indeed, one is characterized by focal adenomatous hyperplasia often macroscopically invisible, whereas the other consists of a diffuse, but discreet, beta-cell abnormality. We evaluated, in a large series of persistent hyperinsulinemic hypoglycemia, the reliability of two criteria in differentiating these two forms: the mean beta-cell nuclear radius (MNR) and the beta-cell nuclear crowding, i.e. the number of nuclei per 1000 micron 2 beta-cell (BCNC). The values of the largest MNR and of BCNC in cases bearing a focal lesion (respectively, 3.27 microns +/- 0.25 and 14.62 +/- 1.78) were significantly different from those in the diffuse pathology (4.25 microns +/- 0.43 and 10.00 +/- 1.55). Setting the threshold value of MNR at 3.70 microns and that of BCNC at 12.00 enabled correct classification of 90.9% of the diffuse and 100% of the focal forms. beta-Cell nuclear analysis can thus contribute to a subclassification of the syndrome, not allowed by clinical or biological data. If performed during surgery it could help in determining the extent of pancreatectomy necessary to cure the patient, as the diffuse form, with abnormal nuclei in the whole pancreas, requires subtotal to near-total pancreatectomy, whereas the focal form, devoid of abnormal insular beta-cell nuclei, can be cured by partial pancreatectomy.