Abstract More people are working into older age, raising questions about how many hours they can work before their health becomes compromised. This paper models work-hour tipping points for mental health and vitality among older Australian workers aged 50–70 years. We use longitudinal data from the Household, Income and Labour Dynamics in Australia (HILDA) survey, 2005–2016 (about 44,900 observations), and bootstrapping Three Stage Least Squares (3SLS) estimation techniques to adjust for reverse and reciprocal relationships between wages, work hours and health. Our approach corrects for heteroscedasticity in the system equation error terms, and we estimate models on the relatively healthy older adults who have remained employed into older age. Among these older workers we observe weekly thresholds of 39–40 hours beyond which mental health and vitality decline. This average, however, hides variability in work-hour limits linked to overall health and occupation. Thus, weekly tipping points for blue- and pink-collar jobs are 7–9 hours lower compared to white-collar jobs, and even wider gaps (11 hours) are apparent for workers with poorer physical functioning, which becomes common as people age. Our modelling reveals that age is not the biggest limiting factor for how many hours older adults can work, rather their health and the types of jobs are critical, and likely widen the gap in who ages successfully or not.