Twenty-two fine needle (30 gauge) aspirations were performed in eyes enucleated for the clinical diagnosis of melanoma. Cytologic preparations were evaluated for adequacy of material, and needle tracts were evaluated for tumor implantation. A scleral marking method was used to identify all needle tracts. The number of tumor cells in tracts of direct transscleral aspirates was compared to those in tracts of indirect aspirates that traversed the anterior chamber or vitreous. Cellular material obtained with 30-gauge needles was sufficient for the diagnosis of malignant melanoma in all but one case. While 14 of 21 (67%) of all fine needle aspiration tracts and eight of 15 (53%) of indirect tracts contained tumor cells, the number of tumor cells was less than that associated with tumor growth in experimental models. Indirect aspirate tracts contained significantly fewer cells than tracts of direct aspirates (P less than .001).