The proapoptotic protein tBid is a member of Bcl-2 family, and it plays an important role in apoptosis by inducing mitochondrial outer membrane permeabilization (MOMP) and lysosomal membrane permeabilization (LMP). Previous studies have shown that the mechanism of tBid-dependent MOMP and LMP depends on tBid interacting with membranes. Researchers hold different opinions about whether tBid itself could induce MOMP and LMP. Some of the researchers insist that tBid must trigger other proteins like Bax or Bak inserting into the membrane, and assembly of tBid itself could not form pores large enough to release cytochrome c. Some others think that tBid just like Bax, can permeabilize mitochondrial outer membrane releasing cytochrome c and lysosomal membrane with the leakage of lysosomal cathepsin B. Here, we want to know whether the tBid itself can induce membrane permeabilization in our model system at low concentration. We use 3 ways to observe tBid and membranes interactions. They are confocal imaging of GUVs (giant unilamellar vesicles), traditional single molecular fluorescence assay, and a recently developed approach, single molecular surface-induced fluorescence attenuation (sm-SIFA). So we can obtain information from single vesicle level and single molecule level. At single vesicle level, we can directly find out whether the GUVs are permeabilized and at the same time the shape of the GUVs is changed. At a single molecule level, we can know the properties of one protein. Especially by using the sm-SIFA, we can obtain the insertion depth of exact residue. Combining the results obtained from different ways under the same conditions, we find that tBid itself can induce the model membrane to permeate, releasing the fluorescent molecules, by oligomerization. What is more, we suggest that the mechanism is that in oligomers some tBids can be inserted deep into the membrane although in oligomers not all the proteins have the same insertion depth. It is indicated that the conformations of tBids in oligomers are diversified. We also prove that the ways we use here are efficient. The GUVs and supported lipid bilayers are indeed tenable model systems. Sm-SIFA has a grand future in the study of protein and membrane interactions.