Maintenance of cellular homeostasis and functionality requires rapid translocation of newly synthesized lipids across the membrane leaflets of the endoplasmic reticulum (ER) where majority of cellular lipid biosynthesis takes place. This process is facilitated without the need for ATP by specific membrane proteins—scramblases—a few of which have been very recently identified in the ER. We have previously resolved the structure of the translocon-associated protein (TRAP) bound to the Sec61 translocon, and found this complex to render the membrane locally thinner. We thus hypothesized that the translocon could provide scrambling pathways in the ER membrane. Here, we first observed non-selective lipid scrambling by reconstituted translocon complexes using complementary fluorescence approaches. Our experiments with inhibitors that block the putative Sec61 scrambling site suggest that an additional scrambling pathway is active in the translocon. Our extensive molecular dynamics simulations indicate that this additional pathway could be provided by the trimeric bundle of TRAP subunits. Its crevice rich in polar residues shields lipid head groups as they traverse the membrane via a credit card mechanism. We analyzed the kinetics and thermodynamics of lipid scrambling by both Sec61 and TRAP and demonstrated that local membrane thinning provides a key contribution to scrambling efficiency. Both proteins appear selective towards phosphatidylcholine lipids over phosphatidylethanolamine and phosphatidylserine, though this trend reflects the natural flip-flop tendencies of these lipids in a protein-free membrane. The identified scrambling pathway in the Sec61 translocon is located at the lateral gate region, which is likely either closed or occupied by a nascent polypeptide during protein translocation. Moreover, our simulations suggest Sec61 scrambling to be impeded by physiological salt concentration. Together with the observations with Sec61 inhibitors, this indicates the presence of an alternative scrambling site in the translocon complex, and the metazoan-specific transmembrane helix bundle of TRAPβ, TRAPγ, and TRAPδ seems like a viable candidate with scrambling activity that is insensitive to the translocon functional state and solvent conditions.