A serology panel was used to assess the comparability of results of poliovirus neutralizing antibody assays. Five laboratories from three countries provided results with seven in-house methods. A high degree of within-laboratory reproducibility was seen for all laboratories and methods, with 94% of results for repeat titrations within plus or minus one twofold dilution. Considerable differences in sensitivity between methods were, however, observed. The period of serum/virus incubation prior to inoculation of cells was shown by one laboratory to have a 10-fold effect on titres, and another laboratory showed that use of Sabin or wild-type challenge virus significantly influenced results for types 1 and 3. Expression of results as relative potencies abolished these difference due to method. The difference in sensitivity between laboratories for end-point titres (if one very sensitive method was excluded) was around fourfold for types 1 and 2, and ninefold for type 3. This range increased to 15-fold for type 2; and 20-fold for types 1 and 3 if the most sensitive method was included. Expression of results relative to a standard considerably improved agreement with a residual variation of around fourfold. Therefore, expression of results from serological studies, including vaccine trials, in International Units (i.e. relative to the new International Standard) will assure that comparisons between studies can be made with confidence.