Background This study aimed to identify the trajectories of work ability over 16 years preceding the individual pensionable age and to examine the association with retirement timing.

Methods The study population consisted of 2612 public sector employees from the Finnish Retirement and Aging study and the Finnish Public Sector study. Participants were grouped into ‘no-extension’ (retired at the individual pensionable date or worked no longer than 6 months after that date) and ‘extension’ (worked more than 6 months after individual pensionable age). Trajectories of self-reported work ability score (0–10) in maximum of eight measurement points over 16 years preceding retirement were examined using the group-based latent trajectory analysis. Log-binomial regression was used to analyse the association between trajectory groups and extended employment.

Results Four stable (‘Stable excellent’, 7%; ‘Stable high’, 62%; ‘Stable medium’, 24%; ‘Low’, 4%) and one decreasing (‘Declining’, 3%) work ability trajectories were identified. After taking into account gender, age, occupational status, marital status and self-rated health, ‘Stable excellent’ trajectory was associated with a higher likelihood of extended employment compared with the ‘Low’ (risk ratio (RR) 2.38, 95% CI 1.21 to 4.68) and to the ‘Declining’ (RR 2.82, 95% CI 1.32 to 6.01) trajectories. There was no difference in retirement timing between ‘Declining’, ‘Low’ and ‘Stable medium’ trajectories.

Conclusion Work ability remained relatively stable among majority of the participants over 16 years of follow-up. Stable excellent work ability from mid-life to late career was associated with higher likelihood of extending employment beyond individual pensionable age than those with low or declining work ability.