Credit registers collect, store and share data on borrowers’ past and current credit relations. Interestingly, the data are typically erased from the public records after a certain number of years, in accordance with privacy protection laws designed to enable people to make a fresh start. However, in order to give creditworthy but unlucky borrowers the chance for a new beginning, these provisions ultimately remove all the public information, including data that may still be relevant for purposes of screening. This paper assesses this trade-off, examining the impact of limited records on borrowers’ behaviour and market outcomes in a stylized credit market for unsecured loans. In this setup, limited records endogenously produce beneficial reputation effects in the form of greater effort in equilibrium. That is, alleviate the limitations rather than worsen the distortions due to asymmetric information. Further, we show that when moral hazard is great, one-period records can increase welfare and lower the default rate by comparison with records that show either all of the past history or none.