Role innovation, the introduction of significant new behaviours into a role, has been a topic of interest in three diverse social science literatures, but few studies have been conducted in field settings. A longitudinal study of job change among 1700 male and female British managers is reported in which theoretical predictions about role innovation deriving from Nicholson's (1984) theory of work-role transitions are tested. Self-concepts and role requirements are examined as predictors of change in reported role innovation following work-role transitions. The relationships between role innovation and reported post-transition satisfaction and personal change are also examined. Job discretion and growth needs emerge consistently as predictors along with previous role innovation. The association between role innovation, satisfaction and personal change suggests that opportunities to role innovate contribute to psychological well-being at work and are exploited by those individuals who are also able to adapt themselves to their environments. The findings imply the need for modification of existing theory by incorporating social and motivational factors and suggest the value of typologies of role innovation.