The membrane-associated RING-CH (MARCH) family of proteins are membrane-embedded E3 ubiquitin ligases that regulate the cell surface expression of substrates via ubiquitin-mediated endocytosis and lysosomal degradation. Although a diverse pool of cell surface proteins have been identified as MARCH substrates, including MHC- class I and class II molecules, the T cell co-receptor CD4, the NK cell ligand Mult1, and leukocyte adhesion receptors, the mechanisms by which MARCH recognises these substrates have not been fully elucidated. A common feature among MARCH proteins is the presence of two transmembrane domains that, in addition to anchoring them to the membrane, have been shown to contribute to substrate preferences. However, no ‘recognition motifs’ within the TM domains of MARCH or substrate have been identified, and the determinants of substrate specificity have yet to be defined. This study aims to characterise the structural and molecular features of the two TM domains of the mammalian MARCH9, and determine the sequences required for substrate recognition. Employing a combination of biophysical and biochemical techniques, such as nuclear magnetic resonance (NMR) spectroscopy and substrate down-regulation assays in cells, we defined the limits of α-helical secondary structure in the TM domains of MARCH9, and identified a key serine residue in the first TM domain that is absolutely required for MARCH9 activity against two different substrates. This provides some of the most compelling evidence to date that specific sequences in the MARCH TM domains are intimately involved in substrate recognition, and uncovering the mechanisms by which TM mutants malfunction will provide new insights into how MARCH proteins identify target molecules.