We have recently shown that the carbohydrate-binding pattern of galectins in cells differs from that determined in artificial (non-cellular) test-systems. To understand the observed discrepancy, we compared several test-systems differing in the mode of galectin presentation on solid phase. The most representative system was an assay where the binding of galectin (human galectins-1 and -3 were studied) to asialofetuin immobilized on solid phase was inhibited by polyacrylamide glycoconjugates, Glyc-PAA. This approach permits us to range quantitatively glycans (Glyc) by their affinity to galectin, i.e. to study both high and low affinity ligands. Our attempts to imitate the cell system by solid-phase assay were not successful. In the cell system galectin binds glycoconjugates by one carbohydrate-recognizing domain (CRD), and after that the binding to the remaining non-bound CRD is studied by means of fluorescein-labeled Glyc-PAA. In an "imitation" variant when galectins are loaded on adsorbed asialofetuin or Glyc-PAA followed by revealing of binding by the second Glyc-PAA, the interaction was not observed or glycans were ordered poorly, unlike in the inhibitory assay. When galectins were adsorbed on corresponding antibodies (when all CRDs were free for recognition by carbohydrate), a good concentration dependence was observed and patterns of specificities were similar (though not identical) for the two methods; notably, this system does not reflect the situation in the cell. Besides the above-mentioned, other variants of solid-phase analysis of galectin specificity were tested. The results elucidate the mechanism and consequence of galectin CRD cis-masking on cell surface.