While a large share of Disability Insurance recipients are expected to recover, outflow rates from temporary disability schemes are typically negligible. We estimate the disincentive effects of disability benefits on the response to a (mental) health improvement using administrative data on all Dutch disability benefit applicants. We compare those below the DI eligibility threshold with those above and find that disincentives significantly reduce work resumption after health improves. Approximately half of the response to recovery is offset by benefits. Structural labor supply model estimates suggest disincentive effects are substantially larger when the workers earnings capacity is fully restored.