The effect of two anti-inflammatory drugs on the development and persistence of clinical signs in cattle experimentally infected with bovine ephemeral fever (BEF) virus was investigated by their administration, either before or after the commencement of fever. A total of 16 cattle was given phenylbutazone sodium (PBZ). The drug prevented fever and other clinical signs in six cattle when given daily during the incubation period, and at 8-h intervals for 5 days when clinical disease might be expected. When treatment with PBZ was deferred until 2–4 h after the commencement o of fever, the rectal temperature returned to normal within 4 h in four of six cattle and the development of other clinical signs was suppressed. Clinical signs of ephemeral fever occurred in four untreated cattle infected at the same time. Viraemia, the development of neutralizing antibodies (at 8–11 days), resistance to subsequent challenge with BEF virus, neutrophilia, lymphopenia and a rise in plasma fibrinogen occurred in all BEF-infected animals whether treated or untreated, despite different clinical appearances. The mean peak of plasma fibrinogen in the untreated cattle was 6.9 g l −1; 3.2 g l −1 when treated 2–4 h after fever developed and 3.8 g l −1 when treated from 18-h post-infection. BEF virus was isolated from leucocytes of each of the cattle, but the frequency of isolation was lower in the treated group. The results indicate that treatment with PBZ blocked the host response which produces the clinical signs and did not have an anti-viral effect. In a similar experiment, a long-acting anti-inflammatory drug, flunixin meglumine, failed to prevent BEF or to modify the clinical signs once they had developed, except for the rectal temperature which returned to normal within 2–4 h of the administration of the drug. The efficacy of this drug was not improved by increasing the dosage to two or three times the recommended level.