Media bias and reputation are central to understanding how news organizations shape their reporting to build a reputation for accuracy. This paper presents a model where media firms slant their reports to align with consumers' prior beliefs, aiming to build a reputation for quality. Even though this can harm all market participants, bias emerges because firms are motivated by reputation rather than truth. The model predicts that bias is less severe when consumers have access to independent evidence about the true state of the world and that competition among independently owned news outlets can reduce bias. Empirical evidence supports these predictions. For example, in contexts where outcomes are easily observable, such as weather forecasting, bias is less prevalent. In contrast, coverage of foreign wars or discussions of tax policy, where outcomes are harder to observe, tends to exhibit more bias. The paper also shows that competition in the news market can lead to lower bias, as firms are more likely to be exposed if they distort their reports. Additionally, when all firms in a market are jointly owned, bias may remain unchanged even as the number of firms increases. The analysis highlights the role of feedback in limiting bias. Rapid feedback, such as in local sports reporting, helps reduce the incentive to slant information. The paper also discusses how media firms in more competitive markets have stronger incentives to reveal important information, as evidenced by the more equitable treatment of candidates in television news reports leading up to the 2000 election. The model is related to the literature on "herding on the priors," where agents' incentives to act on information depend on the prior beliefs of those who determine their rewards. The paper also connects to work on reputational effects in sender-receiver games, emphasizing the importance of the receiver's prior beliefs in shaping reporting strategies. The findings suggest that bias can arise even when news consumers care only about learning the truth, and news sellers care only about maximizing profits. The model predicts that increased competition reduces the incentive to bias reports toward consumer priors, contrasting with some existing theories. The paper concludes that media bias is influenced by the desire to build a reputation for accuracy, and that competition and feedback mechanisms can reduce this bias. Empirical evidence supports these findings, showing that media firms in competitive markets are more likely to report information fairly and accurately.Media bias and reputation are central to understanding how news organizations shape their reporting to build a reputation for accuracy. This paper presents a model where media firms slant their reports to align with consumers' prior beliefs, aiming to build a reputation for quality. Even though this can harm all market participants, bias emerges because firms are motivated by reputation rather than truth. The model predicts that bias is less severe when consumers have access to independent evidence about the true state of the world and that competition among independently owned news outlets can reduce bias. Empirical evidence supports these predictions. For example, in contexts where outcomes are easily observable, such as weather forecasting, bias is less prevalent. In contrast, coverage of foreign wars or discussions of tax policy, where outcomes are harder to observe, tends to exhibit more bias. The paper also shows that competition in the news market can lead to lower bias, as firms are more likely to be exposed if they distort their reports. Additionally, when all firms in a market are jointly owned, bias may remain unchanged even as the number of firms increases. The analysis highlights the role of feedback in limiting bias. Rapid feedback, such as in local sports reporting, helps reduce the incentive to slant information. The paper also discusses how media firms in more competitive markets have stronger incentives to reveal important information, as evidenced by the more equitable treatment of candidates in television news reports leading up to the 2000 election. The model is related to the literature on "herding on the priors," where agents' incentives to act on information depend on the prior beliefs of those who determine their rewards. The paper also connects to work on reputational effects in sender-receiver games, emphasizing the importance of the receiver's prior beliefs in shaping reporting strategies. The findings suggest that bias can arise even when news consumers care only about learning the truth, and news sellers care only about maximizing profits. The model predicts that increased competition reduces the incentive to bias reports toward consumer priors, contrasting with some existing theories. The paper concludes that media bias is influenced by the desire to build a reputation for accuracy, and that competition and feedback mechanisms can reduce this bias. Empirical evidence supports these findings, showing that media firms in competitive markets are more likely to report information fairly and accurately.