Most integral membrane proteins located within the endomembrane system of eukaryotic cells are first assembled co-translationally into the endoplasmic reticulum (ER) before being sorted and trafficked to other organelles. The assembly of membrane proteins is mediated by the ER translocon, which allows passage of lumenal domains through and lateral integration of transmembrane (TM) domains into the ER membrane. It may be convenient to imagine multi-TM domain containing membrane proteins being assembled by inserting their first TM domain in the correct orientation, with subsequent TM domains inserting with alternating orientations. However a simple threading model of assembly, with sequential insertion of one TM domain into the membrane after another, does not universally stand up to scrutiny. In this article we review some of the literature illustrating the complexities of membrane protein assembly. We also present our own thoughts on aspects that we feel are poorly understood. In short we hope to convince the readers that threading of membrane proteins into membranes is 'not sew simple' and a topic that requires further investigation.