One of the oldest and most important questions in cancer immunology concerns the existence of tumor-specific antigens (1). The existence of such antigens would represent a qualitative difference between a malignant and a normal cell, and the antigen would be clinically useful in tumor diagnosis, prophylaxis, and therapy. The nature of tumor antigens might also give us a clue to the causes of the malignant process itself. However, despite vast efforts by immunologists during the last century, the existence of tumor-specific antigens in the majority of cancers remains unproven. This absence of proof despite extensive effort has commonly led to the view that tumor-specific antigens simply do not exist. However, critics usually underestimate the considerable technical difficulties and pitfalls of experi ments designed to prove the existence of tumor-specific antigens, especially on human cancer cells. In this review, therefore, we try to evaluate critically the available evidence for the existence of tumor-specific antigens, and we also discuss the major problems that have hampered rapid progress in this important field of tumor immunology.