Animal models of impaired sensorimotor gating, as assessed by prepulse inhibition (PPI) of startle, have demonstrated clear validity at face, predictive, and construct levels for schizophrenia (SZ) therapeutics, neurophysiological endophenotypes, and potential causative insults for this group of disorders. However, with the growing recognition of the heterogeneity of the schizophrenias, and the less sanguine view of the clinical value of antipsychotic (AP) medications, our field must look beyond "validity," to assess the actual utility of these models. At a substantial cost in terms of research support and intellectual capital, what has come from these models, that we can say has actually helped schizophrenia patients? Such introspection is timely, as we are reassessing not only our view of the genetic and pathophysiological diversity of these disorders, but also the predominant strategies for SZ therapeutics; indeed, our field is gaining awareness that we must move away from a "find what's broke and fix it" approach, toward identifying spared neural and cognitive function in SZ patients, and matching these residual neural assets with learning-based therapies. Perhaps, construct-valid models that identify evidence of "spared function" in neural substrates might reveal opportunities for future therapeutics and allow us to study these substrates at a mechanistic level to maximize opportunities for neuroplasticity. Such an effort will require a retooling of our models, and more importantly, a re-evaluation of their utility. For animal models to remain relevant in the search for schizophrenia therapeutics, they will need to focus less on what is valid and focus more on what is useful.