The acute phase protein, alpha1-acid glycoprotein (AGP), is a normal constituent of human blood (0.2-1 mg ml(-1)) and its glycosylation and concentration in the blood change during inflammation. In this review of our recent work, we discuss the immunomodulatory properties of AGP in connection with the structure of its carbohydrate chains. AGP samples prepared from normal donor serum (nAGP), serum obtained during abortion (fAGP), serum of cancer patients (cAGP), and ascitic fluid of patients with stomach cancer (sAGP) were subjected to analysis. All the samples except for fAGP had five N-linked chains of the 'complex' type, however, the numbers of bi-, tri-, and tetra-antennary chains, as well as glycan structures terminating these chains, were different. fAGP had three N-linked chains of the lactosamine and polylactosamine type and three O-chains which were not present in AGP isolated from the other sources. The glycoforms of nAGP and sAGP that were isolated using a ConA affinity column were similar in respect to their branching, but differed in their terminal oligosaccharides. sAGP was enriched in units ending in Le(x) and asialoagalacto (GlcNAc-terminating) forms. Immunomodulatory activity of different AGP preparations was tested in vitro by measuring their effect on the proliferative response of human lymphocytes stimulated by PHA, and by determining their influence on the production of IL-1, IL-2, IL-6, and TNF in the stimulated cells. nAGP was less active compared to cancer or fetal AGP in the proliferation test, but more active in affecting cytokine production. Some AGP glycoforms had opposite immunomodulatory effects. A new approach was developed in order to clarify the role of carbohydrate chains in the biological activity of AGP. A pool of N-linked oligosaccharide chains were attached to a soluble polyacrylamide matrix. This 'pseudoglycoprotein' was similar to AGP in its molecular weight; in its relative amounts of tetra-, tri-, and bi-antennary chains; and in the content of mono-, di-, tri-, and tetra-sialylated-oligosaccharides. This pseudo-AGP displayed a similar activity to its parent AGP in the biological tests. Analytical flow cytometry of leukocyte subpopulation from human peripheral blood showed that monocytes and granulocytes but not lymphocytes were the main targets for the binding of AGP and pseudo-AGP. This binding was inhibited by synthetic glycoconjugates containing mannose or sialic acid. The binding curve data suggested that there are two monocyte and granulocyte populations. These may have different carbohydrate specificities. All the evidence provided by these studies indicate that it is the carbohydrate chains on AGP that are important in modulating the immune system and not the AGP molecule itself.