Biomarkers are characteristics that can be measured as an indicator of a normal biological process, and they have special relevance in Parkinson's disease. Parkinson's disease is a chronic neurodegenerative disorder that is difficult to study, given the site of pathology and because the resultant clinical phenotype fluctuates over time. We currently have no definitive diagnostic test, and thus for the clinician there is hope that biomarkers will help diagnose symptomatic and presymptomatic disease or provide surrogate end-points to demonstrate clinical efficacy of new treatments, such as neuroprotective therapies, and help stratify this heterogeneous disease. No biomarker is likely to fulfil all these functions, so we need to know how each has been validated in order to understand their uses and limitations, and be aware of potential pitfalls. In this review we discuss the current potential biomarkers for Parkinson's disease, highlight the problems with their use, and conclude with a discussion of future alternatives.