Increasing the pension age as a dominant solution to population ageing does not bring desirable outcomes, if not accompanied by other essential measures in lifelong learning and fighting age discrimination. Moreover, rapid digitalisation and automation in the labour market bring additional uncertainties for the growing group of older workers. The analysis is based on the SHARE data from Waves 5, 6, and 7 and examines predictors of retirement intentions by two different estimation methods. While digital skills are positively associated with a willingness to stay in the labour market in the random-effect modelling, fixed-effects regression shows no correlation between digital skills and retirement intentions. This difference means that digital skills do not correlate with retirement intentions once we control for time-invariant individual characteristics. Thus, increasing ICT literacy among older workers can have a very limited potential for extending working lives. In contrast to this result, starting to be self-employed, health improvement, having an additional grandchild, and losing a partner increase the willingness to work longer. The study identifies the factors shaping retirement intentions, which should be reflected in any effective social policy.