Objectives Cerebrospinal fluid α-synuclein (CSF α-syn) represents a possible biomarker in Parkinson’s disease (PD) diagnosis. CSF blood contamination can introduce a bias in α-syn measurement. To date, CSF samples with a red blood cells (RBC) count >50 RBC × 10 6 /L or haemoglobin (Hb) concentration >200 μg/L are excluded from biomarker studies. However, investigations for defining reliable cut-off values are missing. Methods We evaluated the effect of blood contamination on CSF α-syn measurement by a systematic approach in a cohort of 42 patients with different neurological conditions who underwent lumbar puncture (LP) for diagnostic reasons. CSF samples were spiked with whole blood and serially diluted to 800, 400, 200, 100, 75, 50, 25, 5, 0 RBC × 10 6 /L. CSF α-syn and Hb levels were measured by ELISA. Results In neat CSF, the average concentration of α-syn was 1,936 ± 636 ng/L. This value increased gradually in spiked CSF samples, up to 4,817 ± 1,456 ng/L (+149% α-syn variation) in samples with 800 RBC × 10 6 /L. We established different cut-offs for discriminating samples with α-syn level above 5, 10, and 20% variation, corresponding to a Hb (RBC) concentration of 1,569 μg/L (37 RBC × 10 6 /L), 2,082 μg/L (62 RBC × 10 6 /L), and 3,118 μg/L (87 RBC × 10 6 /L), respectively. Conclusions Our data show the high impact of CSF blood contamination on CSF α-syn levels, highlighting the measurement of Hb concentration as mandatory when assessing CSF α-syn. The thresholds we calculated are useful to classify CSF samples for blood contamination, considering as reliable only those showing a Hb concentration <1,569 μg/L.