The need for an early and differential diagnosis of Parkinson's disease (PD) is undoubtedly one of the main quests of the century. An early biomarker would enable therapy to begin sooner and would, hopefully, slow or better prevent progression of the disease. We performed transcript profiling via quantitative RT-PCR in RNA originating from peripheral blood samples. The groups were de novo (n = 11) and medicated PD (n = 94) subjects and healthy controls (n = 34), while for negative control Alzheimer's disease (AD; n = 14) subjects were recruited as an additional neurodegenerative disease. The results were retested on a second recruitment consisting 22 medicated PD subjects versus 33 controls and 12 AD. Twelve transcripts were chosen as candidate genes, according to previous postmortem brain profiling. Multiple analyses resulted in four significant genes: proteasome (prosome, macropain) subunit-alpha type-2 (PSMA2; p = 0.0002, OR = 1.15 95% CI 1.07-1.24), laminin, beta-2 (laminin S) (LAMB2; p = 0.0078, OR = 2.26 95% CI 1.24-4.14), aldehyde dehydrogenase 1 family-member A1 (ALDH1A1; p = 0.016, OR = 1.05 95% CI 1.01-1.1), and histone cluster-1 H3e (HIST1H3E; p = 0.03, OR = 0.975 95% CI 0.953-0.998) differentiating between medicated PD subjects versus controls. Using these four biomarkers for PD diagnosis, we achieved sensitivity and specificity of more than 80%. These biomarkers might be specific for PD diagnosis, since in AD subjects no significant results were observed. In the second validation, three genes (PSMA2, LAMB2 and ALDH1A1) demonstrated high reproducibility. This result supports previous studies of gene expression profiling and may facilitate the development of biomarkers for early diagnosis of PD.