The use of the mouse lethality assay for the estimation of the biologic activity of botulinum toxin was evaluated. The relationship between the number of animals, number of doses, and duration of the assay used to estimate the LD50 and the precision of the assay was investigated. The results of these studies demonstrated that the LD50 for botulinum toxin can be estimated with a high degree of precision (±5%). The precision of the assay is not increased by using more than a 5-dose 50-animal assay or extending the duration of the assay beyond 72 hr. Estimates of the LD50 obtained at 48 hr were only slightly less precise but underestimated the LD50 by 15%. Analysis of the commercially available preparations of botulinum toxin with the mouse LD50 assay revealed significant discrepencies between the units of toxin in these preparations. In addition, a 2.67-fold difference in the relative potency of the two preparations of botulinum A toxin was observed using a regional chemodenervation assay that measures paralysis. The mouse LD50 assay could not detect this large difference in the potency of the two approved clinical preparations of botulinum toxin. The results of these studies demonstrate that although the mouse LD50 assay can be used to estimate the number of units of botulinum toxin with a high degree of precision this assay alone is not an adequate method for assessing the preclinical biological potency of botulinum toxin.