Neuropeptide S (NPS) is a factor associated with the central regulation of body weight, stress, anxiety, learning, memory consolidation, wakefulness–sleep cycle, and anti-inflammatory and neuroplastic effects. Its stress-reducing, anti-anxiety, arousal without anxiety, and pro-cognitive effects represent an interesting option for the treatment of neuropsychiatric disorders. The purpose of the study was to examine the potential associations of NPS levels in the blood with clinical and metabolic parameters during the rehabilitation therapy of patients with schizophrenia. Thirty-three male subjects diagnosed with schizophrenia were randomly divided into two groups. The rehabilitation group (REH, N16) consisted of patients who were subjected to structured, 3-month intensive rehabilitation therapy, and the control group (CON, N17) consisted of patients who were subjected to a standard support mechanism. Both groups continued their pharmacological treatment as usual. The NPS concentration, as well as clinical and metabolic parameters, were compared in both groups. Additionally, a group of healthy (H) males (N15) was tested for NPS reference scores. To look for the specificity and selectivity of the NPS relationship with clinical results, various factor models of the positive and negative syndrome scale (PANSS) were analyzed, including the original PANSS 2/3 model, its modified four-factor version, the male-specific four-factor model, and two five-factorial models validated in large groups in clinical and multi-ethnic studies. Results and conclusions: (1) Structured rehabilitation therapy, compared to unstructured supportive therapy, significantly reduced the level of schizophrenia disorders defined by various factor models derived from PANSS. (2) The clinical improvement within the 3-month rehabilitation therapy course was correlated with a significant decrease in neuropeptide S (NPS) serum level. (3) The excitement/Hostility (E/H) factor, which included schizophrenic symptoms of the psychotic disorganization, was specific and selective for the reduction in serum NPS, which was stable across all analyzed factor models. (4) The long-term relationship between serum NPS and clinical factors was not accompanied by basic metabolic parameters.