The academic interest in smart working, a form of flexible work characterized by the use of technology to conduct one’s work, has dramatically increased over recent years, especially during the COVID-19 pandemic. Building on the job demands–resources (JD-R) model, in this study we investigate whether smart working affects the longitudinal association between perceived work characteristics, such as workload and social support (SS), and workers’ health and well-being, in terms of exhaustion. Overall, 185 workers completed a self-report questionnaire at two time points (four-month time-lag) during the COVID-19 outbreak. The results from moderated multiple regression analysis partially support our predictions. The longitudinal association between workload and exhaustion was positive—although marginally significant—for smart workers, but nonsignificant for in-person workers. Contrarily, the longitudinal association between SS and exhaustion was negative for in-person workers, but nonsignificant for smart workers. Overall, this study suggests that, to support employees’ health and productivity, work characteristics—both physical and psychosocial—should fit the new way of working as well as remote workers’ specific needs and expectations. Hence, to promote sustainable work, interventions should be aimed at helping smart workers to manage their workload effectively, as well as reducing professional and social isolation.