The complement fixation (CF) test commonly is used to identify cattle infected with Anaplasma marginale prior to interstate or international movement. Estimates of the accuracy of the CF test in detecting animals persistently infected with A. marginale vary widely. In this study, the sensitivity and specificity of the CF test for detection of carrier animals was determined using serum from 232 cattle previously defined as A. marginale positive or negative by nested polymerase chain reaction methods and hybridization. Considering results from 2 independent laboratories and interpreting a 1:5 suspect reaction as positive, the best estimate of CF test sensitivity was 20%, with a specificity of 98%. Using a 1:10 cutoff, sensitivity decreased to 14% and specificity increased to 99%. Results of this study indicate that the CF test is ineffective for identifying cattle persistently infected with A. marginale and thus is inadequate for anaplasmosis regulatory and surveillance programs.