We have investigated the glycosylation, disulfide bonding, and subunit structure of mouse TRPM8. To do this, amino-terminal c-myc or hemagglutinin epitope-tagged proteins were incorporated and expressed in Chinese hamster ovary cells. These modifications had no obvious effects on channel function in intracellular calcium imaging assays upon application of agonists, icilin or menthol, and cold temperatures. Unmodified TRPM8 migrates with an apparent mass of 129 kDa and can be glycosylated in Chinese hamster ovary cells to give glycoproteins with apparent masses of 136 and 147 kDa. We identified two potential N-linked glycosylation sites in TRPM8 (Asn-821 and Asn-934) and mutated them to show that only the site in the putative pore region at position 934 is modified and that glycosylation of this site is not absolutely necessary for cell surface expression or responsiveness to icilin, menthol, and cool temperatures. Enzymatic cleavage of the carbohydrate chains indicated that they are complex carbohydrate. The glycosylation site is flanked in the pore by two cysteine residues that we mutated, to prove that they are involved in a conserved double cysteine motif, which is essential for channel function. Mutation of either of these cysteines abolishes function and forces the formation of a non-functional complex of the size of a homodimer. The double cysteine mutant is also non-functional. Finally, we showed in Perfluoro-octanoic acid-polyacrylamide gels that TRPM8 can form a tetramer (in addition to dimer and trimer forms), consistent with current thinking that functional TRP ion channels are tetrameric.