Changing trends in US mesothelioma incidence In a letter, Greenberg, commenting on our paper, ‘‘Changing trends in US mesothelioma incidence’’, raises a number of points with which we disagree. There is now a broad consensus that amphiboles are vastly more dangerous than chrysotile in their propensity to produce mesothelioma, and even a casual review of the literature indicates that where there is a continuing increase in mesothelioma rates, it is seen in countries that used large amounts of amosite and crocidolite, as we indicated in our paper. He raises the issue of whether asbestosis must be present to attribute a lung cancer to asbestos exposure. Our paper was not about this issue, we did not say anything about necessary sequence, and the phrase he cited was the lone mention of this issue, included in the introduction before focusing on the main subject: mesothelioma. However, since Greenberg raises the subject, the Wilkinson et al paper was in no way a ‘‘consensus’’ by a ‘‘group of experts’’, rather it was the report of a study of hospitalised patients, a study which had a number of serious flaws, as we explained in our published response. Also, even if one assumes that lung cancers can be generated by high exposure to asbestos without the presence of asbestosis, it makes no sense to expect an increased lung cancer risk in the face of declining asbestos usage and increased control of exposure, exactly the factors that are driving the decreased rates of mesothelioma. Greenberg’s own words clearly indicate that he is more interested in the ‘‘adversarial spectrum’’ than the science! Greenberg complains about the use of national import tonnage as an indicator of potential overall worker exposures to the various asbestos fibre types. There is no other way to do this, and this approach has been used by Peto and colleagues, and in a previous publication of ours, comparing US and UK mesothelioma trends. We do not know exactly what paper Greenberg refers to in his comments about events that occur in test tubes within 4 minutes, but in vitro experimental data must always give way to in vivo experimental data, and the latter to human epidemiology. Greenberg is actually incorrect in stating that all types of fibres are equally potent causes of malignancies in animals. In fact, proper analysis of the original Wagner inhalation experiments in rats indicates that, because the asbestos was delivered on an equal mass and not equal fibre number basis, the number of chrysotile fibres to which the animals were exposed was vastly greater than the number of amphibole fibres but the number of mesotheliomas found was about the same, thus indicating the greater potency of amphiboles in causing mesothelioma. This conclusion has been confirmed for both mesothelioma and lung carcinoma in more modern animal inhalation studies, comprehensively reviewed, and there is in fact considerable evidence that lung cancers in animals exposed to asbestos only develop when asbestosis is present. Greenberg raises a number of other issues, but the overall thrust of his letter appears to be that declining rates of (readily apparent) asbestos induced disease should not be viewed as indicating a foreseeable end to the asbestos problem. Yet if declining rates of disease do not indicate a problem that, with proper control of exposure, will slowly disappear, what will? Finally, Greenberg closes with kudos for a conference entitled ‘‘The third wave of asbestos disease’’. In the view of most investigators in this field, this conference contributed little to the knowledge base on asbestos related diseases and served mainly as a political/litigation brief. One wonders if there will ever come a time when any good news about asbestos related health effects is welcomed by all who profess to have worker health as their primary motivation.