Ten laboratories studied the immunogenicity of six trivalent inactivated poliovirus vaccines in a WHO Collaborative Study. The antigenic content of all six vaccines had previously been estimated in an earlier study (Woodet al., 1995). In collaboration with the European Pharmacopoeia Commission an additional preparation, European Pharmacopoeia Biological Reference Preparation Batch 1, was also included. Laboratories were requested to use established immunogenicity assays and six used the European Pharmocoepia guinea pig/chick test while three used a rat potency test. One laboratory contributed data from both methods. Apart from one laboratory, within laboratory variation was low (less than five-fold). However, very large (greater than 100-fold) variation was seen between laboratories for ED50 results in the guinea pig/chick test. Different decisions on pass/fail outcome would have occurred for some of the samples tested. Between laboratory variation was much lower (less than five-fold) in the rat test. Expression of results as potencies relative to a standard reduced between laboratory variation for both methods, substantially so for the guinea pig/chick test. The correlation betweenin vivoandin vitroresults was generally good with the exception of the type 3 component of one preparation. This showed that the relationship between immunogenicity and antigenicity was not necessarily predictable. There is an urgent need to revise the European Pharmocoepia immunogenicity test but it is premature, on the basis of this study, to recommend either the rat test orin vitrotests as replacements. Two candidate reference materials were both found suitable forin vivoassay of inactivated poliovirus vaccine.