Avian field studies commonly employ colored leg bands to follow individual birds without recapturing them. However, this technique is subject to several types of reporting errors. In a laboratory setting under ideal conditions, we marked model birds with colored band combinations to examine resighting rates and errors made by trained and untrained observers. We varied both the time of exposure and the number of birds presented to observers in a series of replicate trials. We found large variation in the number of incorrectly recorded combinations among both trained and untrained observers. The mean error rate for trained observers was 5%, and for untrained observers was 16%. In both groups, error rates significantly increased when observers were exposed to more birds or observation intervals were short. The most common type of error involved switching band combinations on left and right legs. In some cases incorrectly recorded combinations matched actual combinations used on other birds. We also found that resighting probabilities depended on the particular colors used. Because this study was performed under indoor lighting conditions and at a relatively close distance (to mimic conditions under which birds are viewed at nearby feeders), we suggest that the error rates we observed likely represent low estimates and that error rates under field conditions will be higher.