Why does the same monetary tightening lead some banks to write safer loans than others? We develop a model and identify a novel deposit-franchise mechanism within the risk-taking channel of monetary policy. Sticky deposits generate rents that vanish if the bank fails, giving low-deposit-beta banks more skin in the game. The model predicts that tightening lowers the payoff to risk-taking, so banks with a stronger deposit franchise cut risk more because failure destroys larger rents. We test this prediction using the Federal Reserve’s confidential loan-level data, interacting high-frequency monetary policy surprises with predetermined deposit betas in specifications with bank and borrower-time fixed effects. Our findings show that monetary tightening reduces risk-taking, especially at low-beta banks, and the result survives a horse race with bank capital. New-loan originations reveal how banks de-risk; for the same borrower-quarter, low-beta banks originate loans that are more likely to be collateralised and senior.