Background: Phase angle measured by bioelectrical impedance analysis (BIA) may be a marker of health state. Objective: This historical cohort study of prospectively collected BIA measurements aims to investigate the link between phase angle and mortality in older people and evaluate whether a phase angle cutoff can be defined. Design: We included all adults aged ≥65 years who underwent a BIA measurement by the Nutriguard ^® device at the Geneva University Hospitals. We retrieved retrospectively the phase angle and comorbidities at the last BIA measurement and mortality until December 2012. We calculated phase angle standardized for sex, age, and body mass index (BMI), using reference values determined with the same brand of BIA device. Sex-specific and standardized phase angle were categorized into quartiles. The association of mortality with sex-specific or standardized phase angle was evaluated through univariate and multivariate Cox regression models, Kaplan–Meier curves, and receiver operating characteristic (ROC) curves. Results: We included 1307 (38% women) participants, among whom 628 (44% women) died. In a multivariate Cox regression model adjusted for comorbidities and setting of measurement (ambulatory vs. hospitalized), the protective effect against mortality increased progressively as the standardized phase angle quartile increased (HR 0.71 [95% CI 0.58, 0.86], 0.53 [95% CI 0.42, 0.67], and 0.32 [95% CI 0.23, 0.43]). The discriminative value of continuous standardized phase angle, assessed as the area under the ROC curve, was 0.72 (95%CI 0.70, 0.75). We could not define an acceptable phase angle cutoff for individual prediction of mortality (LK), based on sensibility and specificity values. Conclusions: This study shows the association of phase angle and mortality in older patients, independent of age, sex, comorbidities, BMI categories, and setting of measurement.