An important part of the field of experimental pharmacology encompasses the study of the effects of molecules in animals. In the case of Parkinson's disease (PD), animal models have played an invaluable role in refining our understanding of the disease, in characterizing the effect of new compounds on the illness, and in bringing new treatments to the clinic. Indeed, it is now hardly conceivable that a drug would be tested in humans if several parameters pertaining to safety, toxicology, efficacy, etc. had not previously been evaluated in animal models. Quite unfortunately, efficacy in experimental models of PD does not necessarily guarantee positive clinical outcomes. In this article, which primarily focusses on the past five years, we review drugs that entered clinical trials for the treatment of levodopa-induced dyskinesia, parkinsonism, disease modification, and had previously been assessed in animal models of PD. The drugs discussed are buspirone, JM-010, befiradol, mesdopetam, foliglurax, dipraglurant, tavapadon, prasinezumab, cinpanemab, nilotinib, minzasolmin, exenatide, NLY01, liraglutide, lixisenatide, and semaglutide. For each molecule, we examine how previous preclinical studies succeeded or failed in predicting efficacy in clinical trials and discuss possible ways to optimize animal-model design and selection to maximize the probability of translational success.