Cannabinoid (CB1) receptor is a therapeutic drug target, and its structure and conformational changes after ligand binding are of great interest. To study the protein conformations in ligand bound state and assist in drug discovery, CB1 receptor homology models are needed for computer-based ligand screening. The known CB1 ligands are highly diverse structurally, so CB1 receptor may undergo considerable conformational changes to accept different ligands, which is challenging for molecular docking methods. To account for the flexibility of CB1 receptor, we constructed four CB1 receptor models based on four structurally distinct ligands, HU-210, ACEA, WIN55212-2 and SR141716A, using the newest X-ray crystal structures of human β₂ adrenergic receptor and adenosine A(2A) receptor as templates. The conformations of these four CB1-ligand complexes were optimized by molecular dynamics (MD) simulations. The models revealed interactions between CB1 receptor and known binders suggested by experiments and could successfully discriminate known ligands and non-binders in our docking assays. MD simulations were used to study the most flexible ligand, ACEA, in its free and bound states to investigate structural mobility achieved by the rearrangement of the fatty acid chain. Our models may capture important conformational changes of CB1 receptor to help improve accuracy in future CB1 drug screening.