It is difficult to identify the bioactive conformations of flexible endogenous ligands that have a large number of rotatable bonds, but one approach is to introduce conformational restrictions. Then, by comparing the accessible conformational space and bioactivity of the analogs, it should be possible to estimate the bioactive conformational space. Here, we applied this strategy to identify the bioactive conformation of lysophosphatidylserine (LysoPS(18:1)), an endogenous lipidic ligand of the G-protein-coupled receptor GPR34. We synthesized LysoPS analogues conformationally restricted with respect to the glycerol moiety and sampled the accessible conformation spaces of two active and two inactive analogues by means of long-duration replica exchange molecular dynamics (REMD) and metadynamics (MTD) simulations. The conformational populations of these analogues and LysoPS itself were sampled with respect to two dihedral angles along the glycerol moiety. We found that REMD sampling of the whole molecule and MTD-based estimation of the free energy landscape gave similar results. Furthermore, the dihedral angles of LysoPS(18:1) in the previously identified docking pose with GPR34 were consistent with the range of dihedral angles in bioactive ligand conformations evaluated from the REMD and MTD calculations. Thus, at least with respect to the two dihedral angles of the glycerol moiety, the GPR34-bound conformation of LysoPS(18:1) corresponds well to some of the available unbound ligand conformations. Our findings are consistent with the “conformational selection” model of receptor recognition. This methodology for predicting the bioactive ligand conformation of flexible ligands should be helpful for rational drug design.