Bluetongue virus (BTV) RNA was detected by the polymerase chain reaction (PCR) in the blood of 24 naturally infected cattle as long as 160 days after the estimated date of infection. Blood samples from these animals and from 10 experimentally BTV-infected sheep, which also exhibited a prolonged hematologic BTV RNA presence, were concurrently evaluated for viral infectivity. Infectivity analyses were conducted using the sentinel sheep inoculation and embryonated chicken egg inoculation procedures. Blood specimens from the experimental sheep 50, 56, 71, and 89 days after BTV inoculation were uniformly negative for viral infectivity despite their uniformly positive status with PCR evaluation. Three collections of blood from the naturally infected cattle at least 100, 135, and 160 days after infection also revealed no recoverable viral infectivity but an initially high and progressively decreasing prevalence of BTV with the PCR technique. These retrospective epidemiologic and prospective experimental approaches were concordant in that both studies demonstrated consistent discrepancies between the viral infectivity and the PCR diagnostic data. The significance of these discrepancies is discussed with respect to Koch's postulates and with respect to the possibility that the biological vector of BTV (Culicoides variipennis) may recover BTV infectivity from PCR-positive but virus isolation-negative blood.