Predicting the malignant potential of gastrointestinal stromal tumors (GISTs) remains difficult. We assessed the value of serosal penetration, an established prognostic factor in solid tumors, to determine the clinical outcome in patients with GISTs. From 1996-2002, 25 consecutive patients with GIST underwent surgical resection at our Department. The histopathological presence of serosal penetration was assessed to predict clinical outcome. In addition, the established histopathological classification system by Franquemont (modified by using the Ki-67 proliferation index), was applied to each study patient. A Ki-67 index > or =5% (p<0.001) and a mitotic rate > or =5/50 high-power fields (p<0.047) significantly correlated with a shorter survival, whereas a tumor size >5 cm (p=0.07) tended towards a worse prognosis. The survival of patient groups defined by Franquemont (p=0.03) were of prognostic relevance. The presence of serosal penetration significantly correlated (p<0.01) with a shorter survival. Our data suggest that the presence of serosal penetration is a negative prognostic factor for GISTs. Serosal penetration may become a useful additional parameter for the classification of the malignant potential of GISTs. Since our data are merely hypothesis-generating, serosal penetration should be evaluated in large prospective databases.