AIMS: To evaluate observer variation in diagnosis of villitis of unknown aetiology. METHODS: Fifty haematoxylin and eosin stained sections were circulated to three pathologists who were asked to assess if villitis was present. These slides, with an additional 20, were recirculated and reassessed by the same pathologists. RESULTS: Intra-observer agreement was 84.7% (range 74--92%) and interobserver agreement was 81%. A conjoint review by the three pathologists revealed that sources of differences included the overlooking of isolated single or small numbers of affected villi, the difficulty in assessing stromal cellularity close to infarcted parenchyma, and apparent stromal hypercellularity in immature villi. CONCLUSIONS: Experienced pathologists can show a significant interobserver variation in assessing villitis of unknown aetiology. Future studies on villitis of unknown aetiology should address the problem of observer reproducibility of diagnosis.