Paramyosin is the main structural component of the thick filament of molluscan smooth muscles. These filaments consist of a large paracrystalline core of paramyosin with myosin arranged on its surface. The detailed molecular packing of paramyosin in the core and the array of myosin on the surface of the paramyosin core remain unknown. An unsolved problem is the polarity of the paramyosin molecules within these thick filaments (i.e., it is not known whether the paramyosin molecules assemble with their NH2-terminal ends pointing toward the center or toward the end of the thick filament). Here a method to distinguish between the NH2- and the COOH-terminal ends of the paramyosin molecule by electron microscopy is described and used to determine their polarity in synthetic paracrystalline arrays. This method consists of labeling the cysteine residues of paramyosin molecules with the avidin-biotin system developed by Sutoh et al. (1984). Accordingly, the sulfhydryl groups of paramyosin--isolated from the anterior byssus retractor muscle (ABRM) of Mytilus edulis--were modified with maleimide-biotin, and the biotinylated thiols were visualized in the electron microscope after glycerol spraying/rotary metal shadowing by attaching monomeric avidin to them. Avidin-biotin labeling of the native molecule and its carboxypeptidase fragments revealed that ABRM paramyosin contains one pair of cysteine at its NH2-terminal end and one pair at approximately 30 nm from its COOH-terminal end. Synthetic paracrystalline arrays of paramyosin with known axial arrangement were also labeled with the avidin-biotin system. The location of the bound avidin in these paracrystals indicated the polarity of paramyosin in these arrays. The polarity was also determined by comparison of the transverse band-like staining pattern of paracrystals of alpha-paramyosin (intact protein) and beta-paramyosin (a proteolytically cleaved alpha-paramyosin that has lost a small segment at its COOH-terminal end). Both methods revealed that paramyosin assembles with its NH2-terminal end pointing toward the center of the paracrystals. The implications of this result for the polarity of paramyosin in the native filament core, and for the arrangement of myosin on the surface of molluscan thick filaments, are discussed.