Does probation pay a double dividend? Society saves the cost of incarceration, and convicts preserve their liberty. But does probation also reduce the risk of recidivism? In a meta‐study we show that the field evidence is inconclusive. Moreover, it struggles with an identification problem: those put on probation are less likely to recidivate in the first place. We therefore complement the existing field evidence with a novel lab experiment that isolates the definitional feature of probation: the first sanction is conditional on being sanctioned again during the probation period. We find that probationers are more likely to recidivate (i.e., to reduce their contributions to a joint project), that punishment cost is higher, efficiency lower, and inequity higher. While experimental subjects are on probation, they increase their contributions to a joint project. However, once the probation period expires, they reduce their contributions. While in the aggregate these two effects almost cancel out, critically, those not punished themselves trust the institution less if punishment does not become immediately effective.