A meta-analysis of published randomized studies comparing prophylactic antibiotics to placebo in craniotomies was performed. Ten studies were examined; eight met criteria for inclusion into the meta-analysis. The analysis showed an advantage of antibiotics over placebo at the P < 10-8 level. Tests for homogeneity of effect size between the individual studies showed similar effects of antibiotic treatment between trials, despite variation in the randomization methods and antibiotic regimens used. No statistically significant difference was detected between antibiotic regimens that did or did not cover gram-negative organisms or between single- and multiple-dose regimens. Cumulative meta-analyses showed that this conclusion could have been confidently drawn by 1988, after only four of the eight eligible trials had been published. Trials published since that time have reinforced these conclusions but have not significantly altered them. Future studies should compare proposed new antibiotic regimens with one of those already demonstrated to be effective, not with a placebo.