Bronchopulmonary carcinoid tumors are tumors with a low malignant potential. They metastasize in 5-15% of cases. Accurate histologic preoperative prediction of the presence of regional lymph node metastases is not possible at this time. A retrospective quantitative pathologic analysis was performed to investigate the possibility of predicting the presence or absence of regional lymph node metastases in 24 patients with bronchopulmonary carcinoid tumors. The results of univariate analysis showed that large tumor size was associated significantly more frequently with regional lymph node metastases than small tumor size (P less than 0.01). The other quantitative features, ie, a larger mean nuclear area, higher standard deviation of the nuclear area and the presence of an aneuploid DNA index, were frequently associated with regional lymph node metastases, but this tendency was not significant. In multivariate analysis the combination of tumor size and mean nuclear area predicted the presence or absence of regional lymph node metastases correctly in 80 and 94% of the cases, respectively. These results indicate that the combination of tumor size and mean nuclear area may serve as a guideline to predict the presence of regional lymph node metastases.