Panel data from Australia are used to study the prevalence of work hours mismatch among long hours workers and, more importantly, how that mismatch persists and changes over time, and what factors are associated with these changes. Particular attention is paid to the roles played by household debt, ideal worker characteristics and gender. Both static and dynamic multinomial logit models are estimated, with the dependent variable distinguishing long hours workers from other workers, and within the former, between volunteers, who prefer long hours, and conscripts, who do not. The results suggest that: (i) high levels of debt are mainly associated with conscript status; (ii) ideal worker types can be found among both volunteers and conscripts, but are much more likely to be conscripts; and (iii) women are relatively rare among long hours workers, and especially long hours volunteers, suggesting long hours jobs may be discriminatory. The research highlights the importance of distinguishing conscripts and volunteers to understand the prevalence and dynamics of long work hours.