Objective: To assess the likelihood of astemizole and terfenadine inducing drowsiness compared with placebo in the light of recent reports that these antihistamines are not necessarily drowsiness‐free.Method: Meta‐analysis was used to analyse data obtained from a search of the reports of trials of terfenadine, astemizole and chlorpheniramine. The latter was included as a positive control.Results: Of the 284 trials screened, 229 were unsuitable for detailed analysis for various reasons, including use of concomitant medication, lack of pertinent information and duplicate publication. Fifty five trials contained the necessary details, 18 of these were cross‐over trials and 37 parallel group trials. Our analyses suggest that the design used had an effect on the incidence of drowsiness reported. Both astemizole and terfenadine were no more likely than placebo to induce drowsiness. The Mantel‐Haenszel pooled odds ratio (ORMH) and confidence interval (CI) for terfenadine were ORMH = 1.01; CI = [0.64, 1.57] for the parallel group trials. The astemizole data were too sparse for pooling with largely zero incidence rates. Chlorpheniramine was clearly sedative (ORMH = 4.07; CI = [2.25, 7.36] 95%).In the cross‐over trials, none of the three antihistamines induced more drowsiness than a placebo, thus suggesting that such trials may mask potential subjective adverse effects.Conclusion: Our results are consistent with the hypothesis that astemizole and terfenadine are essentially non‐sedative antihistamine compounds, while chlorpheniramine is sedative. Care should be exercised when evaluating potentially subjective adverse effects and trial design should be considered as a possible confounding factor.