Objective and reliable noninvasive medical diagnostics of a large variety of diseases is still a dream. As a step in the direction of realization, a spectroscopic breath study of cerebral palsy (CP) was performed. Principal component analysis revealed data clustering for a healthy group and CP individuals was observed, with a P-value below 10 −5. Learning algorithms resulted in 91% accuracy in distinguishing the groups. With the help of manual analysis of absorption spectral features of breath samples, two volatile organic compounds were identified that demonstrate significant deviations in the groups. These represent two esters of propionic acid (PPAE). A transportation scheme was hypothesized that links the gut where propionic acid (PPA) and PPAE are produced, the brain of CP patients, through which PPA and PPAE transmit, and the lungs where PPAE releases. The results show a possibility to detect one more brain-related disorder via breath, in this case CP.