With the aim of establishing a versatile and easy synthesis of branched saccharides for biological applications, we used molecular-dynamics simulations to model Lewisy to two classes of di- or triantennary saccharide mimetics. One set of mimetics was based on 1,3,5-tris(hydroxymethyl)cyclohexane (TMC) as the core, the other on furan, and both were derivatised with galactose and/or fucose. The TMC-based saccharides were biotinylated, while the furan disaccharides were treated with maleimide-activated biotin in a Diels-Alder fashion to yield oxazatricyclodecanes (OTDs). These were then assayed as cell-surface labels in human colon (SW480 and CaCo-2), liver (PLC), Glia (U333 CG 343) and ovary (SKOV-3) tumour cell lines. Discrete staining patterns were observed in all cells, usually at one or two poles of the cells, particularly with the asymmetric 3-β-L-fucopyranosyloxymethyl-4-β-D-galactopyranosyloxymethyl-OTD. Normal SV40-transformed fibroblasts (SV80) showed no staining. Adhesion of the highly metastatic mouse melanoma line B16 F10 to fibronectin was inhibited by 80 % by the TMC-digalactoside and by 30 % by 3,4-bis-(β-D-galactopyranosyloxymethyl)furan. None of the saccharide mimetics inhibited the adhesion of the less metastatic B16 F1 line. Migration of B16 F10 cells through MatrigelTM was greatly inhibited by the TMC-digalactoside and weakly inhibited by the TMC-trigalactoside. The saccharide mimetics that had shown the best structural agreements with the terminal saccharides of Lewisy in the molecular dynamics simulation were also the most biologically potent compounds; this underlines the predictive nature of molecular dynamics simulations. The use of the non-saccharide cores enabled us to adapt spacer lengths and terminal saccharides to optimise the structures to bind more avidly to cell-surface lectins.