Fine-needle aspiration (FNA) is a reliable, safe and cost-effective procedure with an established role in the diagnosis of various solid tissue neoplasms. However, the role of FNA in the diagnosis of primary bone tumors, including chondrosarcoma (CS) is controversial. To determine the accuracy of FNA as a diagnostic procedure, the author reviewed the institutional experience of a series of patients with CS who underwent FNA for diagnosis. The author's objectives were to determine the accuracy of the technique as well as possible limitations to sensitivity and specificity, and perhaps to suggest the most appropriate use for this procedure. Computer records and then subsequently archives of the department were searched for patients diagnosed and treated for CS between 1993 and 2003. Patients without adequate clinical follow-up, missing materials or records otherwise unavailable for review were eliminated from study. All patients who underwent FNA for a diagnosis had to have a subsequent histological confirmation to be included in the study. FNAs were largely performed with image-guided assistance. In those that were palpable, the aspiration was performed by the aspiration cytologist using standard methods. Histologic materials were processed according to standard methods. All cytological and histologic materials were reviewed for accuracy and appropriateness of diagnosis by the author. There were 34 aspirates from 32 patients with CS (2 patients with 2 aspirates each). Attempts at diagnoses were made from 27 primary lesions, 6 recurrent lesions, and one metastatic lesion. There were an additional two patients who were assigned a diagnosis of CS on FNA who ultimately were proven to have chondroblastic osteosarcoma. Of the primary CS, 18 were definitively diagnosed as CS or "malignant chondroid neoplasm," 8 of the aspirates were considered equivocal in that an additional diagnostic procedure was required to clarify or confirm the diagnosis. Two aspirates were diagnosed as negative. Both of the false negatives were due to inadequate sampling of the lesion on FNA. Diagnostic accuracy of FNA for primary CS in this series was 67% (18/27). Accuracy for recurrent or metastatic lesions was higher at 86% (6/7). FNA appears to be a reliable means of diagnosis of recurrent and/or metastatic CS in patients with a documented history. In primary lesions, however, the accuracy of the technique is lower. In addition, there are problems of sampling chondroid components of non-CS lesions such as this study's experience with chondroblastic osteosarcoma.