An improved rotary shadowing technique enabled us to visualize chicken gizzard caldesmon (CaD) and its complexes with one or two covalently linked calmodulin (CaM) molecules by electron microscopy. Using a monoclonal antibody against an epitope in the N-terminal region of CaD (anti-N), we can now identify the end of the molecule that is involved in binding to another protein molecule. Thus in the 1:1 complex of CaD and CaM, the CaM molecule was almost always associated with the C-terminus of CaD, indicating preferential CaM-binding to the C-terminal region. We have also studied binding of CaD to filamentous actin (F-actin), using an EM technique that avoids spraying or freeze drying and thereby preserves the structure of F-actin. Only one end of CaD appeared to bind to F-actin, leaving the rest of the molecule projecting away from the filament. While the majority of anti-N bound at the free end of CaD, some antibody molecules were found on F-actin. These findings suggest that either end of CaD can bind to F-actin. Experiments using a monoclonal antibody against the C-terminus of CaD (anti-C) supported this idea. When the native thin filaments that contain endogenous CaD were incubated with anti-N, almost all the bound antibodies were found on the filaments, indicating that the N-terminal regions of CaD interact with actin, and that the binding affinity of the N-terminal region of CaD for actin is higher in vivo than that in vitro, either because the properties of CaD have been altered during purification, or because of the presence of some other component(s) associated with the native filaments.