Parkinson's disease (PD) is a heterogenous multisystem neurodegenerative disorder displaying a marked variation in phenotype and prognosis. Predicting prognosis has important implications for individual prognostication, empowering clinical trials and targeting novel treatments. This study examined which clinical and biochemical factors predicted prognosis in a large cohort of early PD subjects (Discovery cohort, n=945). A systematic review revealed evidence that older age at onset, mild cognitive impairment, verbal fluency, male gender, REM sleep behaviour disorder, and greater axial impairment may predict more motor and cognitive progression. Comparisons between the Discovery cohort and similar published disease and control cohorts established Discovery as being highly representative of the target population. Older age and poor performance on the Purdue assembly, visuo-spatial and verbal fluency tasks at baseline predicted more rapid cognitive decline. Older age, postural instability, pain and anxiety predicted a more rapid functional decline and dependency. Combinations of each of these clinical assessments were specific but not sensitive. Uric acid, vitamin D, epidermal growth factor, apolipoprotein-A1 and C-reactive protein have been cited as the most promising biochemical biomarkers, although there was a lack of high quality studies. In the Discovery cohort, while a lower uric acid level at baseline was found in PD compared to controls, the difference was insufficient to be of diagnostic value. None of the proposed biochemical markers predicted future progression on careful longitudinal follow up. This important negative study should help redirect precious resources to more useful biomarker candidates. In conclusion, a combination of clinic-ready clinical factors at baseline predict a more rapid cognitive and functional decline in PD with good specificity but not sensitivity, and could be used to help tailor prognosis on an individualised basis.