Cattle on Feed (COF) reports are the most important source of cattle supply information for the beef industry. Accuracy of these reports is critical, as they have marked impacts on current and expected fed cattle market prices and influence resource allocation decisions. This study investigates whether COF report revisions are unbiased, random, and anticipated by beef industry analysts. Initial COF reports are biased, but the bias is economically small. Thus, there is no evidence that users of these reports should be concerned about bias. Revisions to COF estimates are not random and exhibit persistence, suggesting when revisions are made subsequent similar types of revisions tend to follow. However, on average, market analysts do not correctly anticipate revisions. Thus, if analysts' prerelease estimates are different from the initial COF report, this does not signal any useful information about future probable COF revisions. [EconLit citations: Q130, D840.] © 2004 Wiley Periodicals, Inc. Agribusiness 20: 363-374, 2004.