Parkinson disease is one of the neurodegenerative diseases that benefited the most from the use of non-human models. Consequently, significant advances have been made in the symptomatic treatments of the motor aspects of the disease. Unfortunately, this translational success has been tempered by the recognition of the debilitating aspect of multiple non-motor symptoms of the illness. Alterations of the sleep/wakefulness behavior experienced as insomnia, excessive daytime sleepiness, sleep/wake cycle fragmentation and REM sleep behavior disorder are among the non-motor symptoms that predate motor alterations and inevitably worsen over disease progression. The absence of adequate humanized animal models with the perfect phenocopy of these sleep alterations contribute undoubtedly to the lack of efficient therapies for these non-motor complications. In the context of developing efficient translational therapies, we provide an overview of the strengths and limitations of the various currently available models to replicate sleep alterations of Parkinson's disease. Our investigation reveals that although these models replicate dopaminergic deficiency and related parkinsonism, they rarely display a combination of sleep fragmentation and excessive daytime sleepiness and never REM sleep behavior disorder. In this light, we critically discuss the construct, face and predictive validities of both rodent and non-human primate animals to model the main sleep abnormalities experienced by patients with PD. We conclude by highlighting the need of integrating a network-based perspective in our modeling approach of such complex syndrome in order to celebrate valid translational models.