The presence of a large number of fibroblast growth factors (FGFs) and multiple splice forms of their receptors (FGFRs) in higher vertebrates makes the three-dimensional (3D) analysis of FGF interactions with their receptors a formidable task. The situation differs in Caenorhabditis elegans (worm) and Drosophila melanogaster (fruit fly), where only one or two FGF and FGFR sequences have been identified. Structural studies of the FGF-FGFR complexes in such primitive organisms should reveal the basic features of the ligand-receptor interactions as they first emerged through evolution. We have analysed the sequences of worm and fly FGFs and FGFRs and used the recently determined crystal structure of the human FGF1-FGFR2-heparin ternary complex [Pellegrini, L., Burke, D.F., von Delft, F., Mulloy, B. and Blundell, T.L. (2000) Nature 407, 1029-34] to construct 3D models of the homologous complexes. In spite of a low sequence similarity with their human counterparts, key structural features required for ligand-receptor and protein-heparin binding in humans are conserved in the fly and worm FGF-FGFR-heparin complexes. Analyses of the models show that tertiary interactions that are not conserved in sequence are maintained through novel interactions or complementary mutations in the fly and worm sequences. The overall charge distributions observed in the human FGF-FGFR-heparin complex are retained in the fly and worm models. The arginine residue at position 253 in the linker region between the Ig-like domains D2 and D3 in the wild type fly and worm sequences is particularly striking, as the Pro253Arg mutation in humans is responsible for Apert syndrome. This change may enhance the affinity of receptors for their FGF molecules as observed in Apert mutants.