We study a model in which an issuer can manipulate information obtained by a credit rating agency (CRA) seeking to screen and rate its financial claim. Better CRA screening leads to a lower probability of obtaining a high rating but makes a high rating more valuable. Over an intermediate range of manipulation cost, improving screening quality can lead to more manipulation, dampening the CRA's incentive to screen. We further show that a CRA's own incentives to inflate ratings constrain its optimal screening intensity. Our model suggests that strategic disclosure by issuers may have played a role in recent ratings failures.