In this paper, we investigate default clusters and reveal credit contagion risk in a data set containing more than 5 million German auto loans.We demonstrate that auto loan defaults cannot be attributed to loan-, borrower- and asset-specific variables and macroeconomic effects alone. Therefore, we explicitly model contagion effects and reveal that the default of one auto loan can lead to the defaults of other auto loans. Our results are highly relevant for banks’ risk management of auto loan portfolios as well as for rating agencies and regulators because they indicate that contagion effects should be considered when assessing credit risk.