Galactosyltransferase (GT) is reported to be a good marker for a number of malignancies. Analytical and preparative attempts to characterize the more specific tumor-associated GT II variant did not yield the expected improved applicability in routine clinical testing for cancer. They indicated instead a puzzling heterogeneity of GT and a special relationship between serum-GT forms and acceptorlike serum glycoproteins. These adducts may represent a general host response to malignant transformation and some nonmalignant events as well. The assay for GT II is, at present, too laborious and time consuming and not readily reproducible. From an extensive discussion of the problems and pitfalls of the GT assay and physiological variation of GT it is concluded that the test is of interest mainly for research purposes and cannot actually serve in the care of patients.