We evaluated the precision of age estimates produced by cementum annuli analysis (CAA) of blind-duplicate specimens taken from 994 southern mule deer (Odocoileus hemionus) collected over 15 years. We found that the mean annual proportion of unreliably aged incisor pairs was greater for females (0.48, SD = 0.13) than for males (0.22, SD = 0.07). Most of the 308 unreliably aged tooth pairs disagreed by only 1 year. Sex, precipitation, and certainty codes assigned by Matson's Lab to the age estimates were the best predictors for agreement of estimated ages within incisor pairs. Our estimated overall age error rate of CAA (17%) was >2 times as large as estimated error rates from Montana and South Dakota, but less than half of error rates estimated for Mississippi and south Texas. Knowing the error rate of age estimates from a specific deer population allows wildlife managers to perform tasks requiring specific age class information such as monitoring the harvest rate of older female deer in a hunted population or performing population reconstruction.