This report describes the development of an HPLC-UV method for studies of glycoamines and glycoamine-like compounds in normal human serum and osteosarcoma patients serum as potential biological markers of cancer. The glycoamines, a newly recognized class of endogenous, low-molecular-mass biopolymers, are conjugates of amino acids and sugar units, containing 5 to 29 amino acid and 1 to 17 sugar units. After ultrafiltration of serum samples, reversed-phase HPLC separation with diode-array detection was used to obtain standard profiles of serum ultrafiltrates below M(r) 10,000 in healthy subjects. These highly reproducible profiles utilized two-dimensional peak identification and were used to develop a statistical profile of the major glycoamine peaks in normal serum. This newly developed analytical method was subsequently used to address a key question: whether or not there is a single tumor-specific glycoamine or a family of tumor-specific glycoamines in cancer patient serum. Preliminary results suggest that this method can separate and detect glycoamines and glycoamine-like compounds in various types of cancer patients serum with a high degree of reproducibility on the basis of comparative two-dimensional identification of natural compounds and a panel of synthetic glycoamine analogs. Moreover, the method is useful for following the relative changes in the amount of a given glycoamine over an extended clinical time course. Initial results suggest that a glycoamine or glycoamine-like compound, GA-4.63, may have clinical utility in human osteosarcoma studies.