Ehrlichia canis, E. equi, and E. risticii seroprevalence was determined by microimmunofluorescent antibody testing (IFA) in a sequential population of 1,845 sick dogs admitted during a 1-year period to the North Carolina State University Veterinary Teaching Hospital. A seroreactor was defined by a reciprocal IFA titer of > or =80 to E. canis, E. equi, or E. risticii antigens. Of the 48 IFA seroreactors, 44 dogs were seroreactive to E. canis, 21 to E. equi, and 0 to E. risticii. Seventeen dogs reacted to both E. canis and E. equi antigens. There was concordance of E. canis IFA and western immunoblot (WI) test results for 36/44 dogs. Because of cross-reactivity of E. canis sera with E. equi antigens, WI was of less utility to confirm E. equi exposure. After elimination of E. canis seroreactors, there was concordance of 2/4 E. equi IFA and WI test results. Based upon a retrospective review of medical records, ehrlichiosis was diagnosed in 10/48 (21%) IFA seroreactive dogs, 9 of which were confirmed positive by WI. Of the remaining 38 IFA seroreactors, 29 also were confirmed by E. canis or E. equi WI. These results indicate that (1) ehrlichiosis was not diagnosed in the majority of serologically confirmed cases, (2) based upon E. canis and E. equi WI analysis, IFA testing was not specific (21% false positive), (3) E. canis sera cross-react with E. equi antigens, and (4) serologic evidence of E. risticii infection was lacking in the dog population studied.