Abstract Background Self-reported low physical activity is a defining feature of phenotypic frailty but does not adequately capture physical activity performed throughout the day. This study examined associations between accelerometer-derived patterns of routine daily physical activity and frailty. Methods Wrist accelerometer and frailty data from 638 participants (mean age 77 [SD = 5.5] years; 44% women) were used to derive 5 physical activity metrics: active minutes/day, sedentary minutes/day, total activity counts/day, activity fragmentation (reciprocal of the average active bout length), and sedentary fragmentation (reciprocal of the average sedentary bout length). Robust, prefrail, and frail statuses were identified using the physical frailty phenotype defined as having 0, 1–2, or ≥3 of the following criterion: weight loss, exhaustion, slowness, self-reported low activity, and weakness. Frailty was collapsed into not frail (robust and prefrail) and frail, and each frailty criteria was dichotomized. Multiple logistic regression was used to model each accelerometer metric. Separate frailty criteria and interactions with age and sex were also examined. Results With higher amounts and intensity of daily activity (more active minutes, fewer sedentary minutes, higher activity counts) and lower activity fragmentation, the odds of frailty were lower compared to robust/prefrail states (p < .02 for all). For interactions, only an age by sedentary fragmentation interaction on the odds of frailty was observed (p = .01). For each separate criteria, accelerometer metrics were associated with odds of slowness, low activity, and weakness. Conclusion Less favorable patterns of objectively measured daily physical activity are associated with frailty and the components of slowness, low self-reported activity, and weakness.