The efficacy of flunixin alone and together with enrofloxacin in treatment of experimental Escherichia coli mastitis was compared using six cows. The cross-over study design was used. Pharmacokinetics of flunixin and enrofloxacin were also studied in these diseased cows. The response of each cow was similar after the first and second challenge and the individual reaction seemed to explain the severity of clinical signs. The most important predictive factor for outcome of E. coli mastitis was a heavy drop in milk yield. Treatment with enrofloxacin and flunixin enhanced elimination of bacteria, but the difference from those receiving flunixin alone was not significant. Two cows, which had received no antimicrobial treatment (Group 1), were killed on day 4 postchallenge. One cow was killed after the first and the other after the second challenge. Cows receiving combination therapy produced 0.9 L more milk per day during the study period than cows which had only received flunixin (P < 0.05). Based on our findings, antimicrobial treatment might be beneficial in the treatment of high-yielding cows in early lactation. The absorption of enrofloxacin was delayed after subcutaneous administration, the mean apparent elimination half-life being about 23 h, whereas after i.v. administration elimination t(1/2) was only 1.5 h. The majority of the antimicrobial activity in milk originated from the active metabolite, ciprofloxacin, which could be measured throughout the 120-h follow-up period after the last subcutaneous administration. No differences were present in the pharmacokinetic parameters of flunixin between treatment groups: mean elimination half-life was 5.7-6.2 h, volume of distribution 0.43-0.49 L/kg and clearance 0.13-0.14 L h/kg. No flunixin or merely traces were detected in milk: one of the three cows had a concentration of 0.019 mg/L 8 h after administration.