OBJECTIVE: To determine whether using intravenous benzylpenicillin in addition to intravenous flucloxacillin would result in a more rapid clinical response in patients with lower limb cellulitis. METHODS: This was a randomised controlled trial set in an inner city teaching hospital, comprising 81 patients with lower limb cellulitis requiring intravenous antibiotics. The main outcome measure was the mean number of doses of antibiotic required until clinical response. RESULTS: The mean number of doses required was 8.47 (95% confidence interval (CI) 7.09 to 9.86) in the benzylpenicillin and flucloxacillin combined group. In the flucloxacillin only group it was 8.71 doses (95% CI 6.90 to 10.5), a mean difference of -0.24 doses (95% CI -2.48 to 2.01, p = 0.83). Other markers of treatment efficacy showed no difference between groups at review the following day; temperature decrease (mean difference -0.07 degrees C, 95% CI -0.76 to 0.62, p = 0.84), or diameter decrease of affected area (mean difference -34 mm, 95% CI -99 to 31, p = 0.30). Patient subjective assessments were also similar between the different drug regimen; improvement on a visual analogue scale of pain/discomfort from admission to first review (mean difference 10 mm, 95% CI -12.6 to 14.2, p = 0.91) and on second review (mean difference 15 mm, 95% CI -18.6 to 21.6, p = 0.88). Patient overall subjective feelings of improvement on first review (p = 0.32) and on second review (p = 0.64) were also similar. CONCLUSIONS: This study provides no evidence to support the addition of intravenous benzylpenicillin to intravenous flucloxacillin in the treatment of lower limb cellulitis.