Purpose Falls are a major problem and cause poor quality of life and reduced life expectancy in people affected by Parkinson's disease (PD). The timed up-and-go (TUG) test is the most popular test of walking ability in patients with PD. However, it carries a risk of falls because patients are told to walk as quickly as possible. The purpose of this study was to characterise issues of safety and reproducibility arising in TUG tests. Methods Our population was 7 hospitalised patients (4 men, 3 women) with PD and 1 healthy control. We used video capture, Motion Analyser software, and an innovative Yorisoi (snuggling nursing robot) robot. Results We found the same movements when patients walked in time to a verbal rhythm provided by the physiotherapist and when they used the robot in walking. The TUG times were longest when the therapist provided no intervention. The typical PD patient experienced freezing of gait when rounding the reflective cone marker in the TUG test, and after turning, experienced gait festination. The patient thus encountered difficulty in using the reflective marker. Moreover, because of the characteristic PD shuffling gait, the patient’s steps were difficult to count automatically. After the turn, the patient's body leaned heavily; however, if the physiotherapist or the robot applied a light force in the direction of the positive axis, it was possible to prevent a fall. Conclusions When applied to patients with PD, the TUG test requires modifications for safety and more accurate step counting to improve reproducibility. IMPLICATIONS FOR REHABILITATION The presence of a sense of security with the attendant or the robot is related to walking speed. The TUG test must be reproducible in patients with PD, regardless of whether there is a rhythm or an attendant, and must be able to count the steps in a shuffling gait. The TUG test must consider safety in patients with PD, because it is easier to fall after a turn and when sitting on a chair, and it is instructed to be in hurry walk. After making a turn, a patient's body may lean heavily; however, if a physiotherapist or a Yorisoi (snuggling nursing robot) robot applies a light force in the direction of the positive axis, it is possible to prevent a fall. Assertive technology, such as the Yorisoi robot, may be used to improve the safety TUG test in patients with PD.