In this article, we estimate the degree of substitution between enrolment into Disability Insurance (DI) and Unemployment Insurance (UI) in the Netherlands. Starting in the 1990s many policy measures aimed at reducing DI enrolment, and increase labour force participation. We quantify whether these policy measures have led to a reduction in hidden unemployment in DI. A side effect of the reforms may be increased pressure on UI. Therefore, we simultaneously estimate reverse substitution, that is, hidden disability in UI. To this end, we employ a sample of firms in the Dutch AVO database from the period 1993 to 2002. Using instrumental variables in a bivariate Tobit specification, we identify the hidden components in both respective schemes. The estimation results indicate that about 3% of all dismissals took place through DI, which implies that about one quarter of the DI enrolments observed in our sample in fact consists of hidden unemployment. We find no evidence for reverse substitution of disabled persons ending up in UI.