Introduction Falls are a frequent and serious complication of Parkinson's disease (PD) in part related to an underlying cholinergic deficit that contributes to both gait and cognitive dysfunction. There is an urgent need to identify strategies that will effectively prevent falls and the consequences thereof. This thesis aims to assess whether ameliorating the cholinergic deficit with the cholinesterase inhibitor rivastigmine will reduce gait variability and the frequency of falls. Methods This phase II randomised double blind placebo controlled trial recruited PD patients, who had fallen in the past year; were able to walk 18 metres without an aid; had no exposure to a cholinesterase inhibitor, and did not have dementia. Patients were randomly assigned (1:1) to rivastigmine (target dose 12mg per day) or placebo by central allocation. The primary outcome measure was step time variability, a marker of gait stability, assessed at 32 weeks employing an intention-to-treat analysis. Step-time variability was assessed in three different walking paradigms combining tasks of increasing attentional demand. Results We recruited 130 patients who were randomly assigned to rivastigmine or placebo and 120 (92.3%) completed the study. Rivastigmine improved step time variability in all three walking conditions with the most significant benefit for normal walking; ratio of geometric means in normal walking 0.72 (95% CI 0.58 to 0.88, p=0.002); simple dual task 0.79 (95% CI 0.62 to 0.99, p=0.05), and complex dual task 0.81 (95% CI 0.60 to 1.09, p=0.17). There was a 45% (95%CI 19% to 62%, p=0.002) reduction in the rate of falls per month during the treatment period. Gastrointestinal side effects were more common on rivastigmine (p