Diseases such as Alzheimer’s cause an alteration of cognitive functions, which can lead to increased daily risk-taking in older adults living at home. The assessment of decision-making abilities is primarily based on clinicians’ global analysis. Usual neuropsychological tests such as the MoCA (Montreal Cognitive Assessment) cover most of the cognitive domains and include mental flexibility tasks. Specific behavioral tasks for risk-taking, such as the Balloon Analogue Risk Task (BART) or the Iowa Gambling Task (IGT), have been developed to assess risk-taking behavior, particularly in the field of addictology. Our cross-sectional study aims to determine whether the MoCA global cognitive assessment could be used as a substitute for behavioral tasks in the assessment of risky behavior. In the current study, 24 patients (age: 82.1 ± 5.9) diagnosed with mild dementia completed the cognitive assessment (MoCA and executive function assessment) and two behavioral risk-taking tasks (BART, simplified version of the IGT). Results revealed no relationship between scores obtained in the MoCA and behavioral decision-making tasks. However, the two tasks assessing risk-taking behavior resulted in concordant risk profiles. In addition, patients with a high risk-taking behavior profile on the BART had better Trail Making Test (TMT) scores and thus retained mental flexibility. These findings suggest that MoCA scores are not representative of risk-taking behavioral inclinations. Thus, additional clinical tests should be used to assess risk-taking behavior in geriatric settings. Executive function measures, such as the TMT, and behavioral laboratory measures, such as the BART, are recommended for this purpose.