My colleagues and I have several concerns with regard to a recent article published in the journal: "lnduction of mesothelioma in p53+/- mouse by intraperitoneal application of multi-wall carbon nanotube" by A. Takagi et al. 33(1):105-116, 2008. While we strongly support the efforts to identify and characterize potential health effects of nanoparticles on workers and consumers as well as evaluation of possible environmental impact, we believe that inherent flaws in the study prevent meaningful hazard identification and risk assessment. First, little information to support risk assessment is gained by using highly unrealistic exposure doses. The authors of this study intraperitoneally instilled 3mg of MWCNT into the abdomen of a mouse. MWCNT are low density materials. Therefore, 3mg represents an inordinately large volume of material. Such exposure is beyond the realm of reality. Therefore, artifactual responses are likely. Secondly, the authors argue that fiber counts for asbestos exposure were an order of magnitude greater than that for the MWCNT exposure. The authors are correct in stating that fiber number is the most appropriate dose metric to use in evaluating the toxicity of fibers. However, the number of fibers per mass for the MWCNT was vastly underestimated. By the authors' own admission, the MWCNT sample was highly agglomerated with structures as large as 50-200 micrometers in diameter. They counted particles not individual nanotubes. A 50 micrometer structure could contain thousands of individual nanotubes. In contrast, asbestos was well dispersed into individual fibers. Therefore, exposure dose expressed as fiber number could be orders of magnitude higher for MWCNT than asbestos in this study. Thirdly, histological evidence shows "typical epithelial mesotheliomas" with "large tumors" after exposure to asbestos. In contrast, MWCNT cause "peritoneal adhesion and fibrous thickening" "due to the formation of fibrous scars" with an associated "spectrum of peritoneal mesothelial lesions". According to the text, it seems that this spectrum of lesions included typical mesotheliomas, but the authors did not present any photomicrograms demonstrating the presence of such mesotheliomas following exposure to MWCNT. As a consequence, the paper did not show any histological evidence that typical mesotheliomas were associated with MWCNT. Fourthly, gross inspection of abdominal structures in the MWCNT -exposed mice showed "moderate to severe fibrous adhesions." The authors report that a "major cause of death was construction ileus due to severe peritoneal adhesion." Therefore, the higher death rate in the MWCNT vs. asbestos group may have been simply strangulation of the intestine; not mesothelioma. Finally, poor material characterization in the study as presented severely compromises interpretation of the results, or comparison with other similar studies. SEM photographs used to determine the size distribution in Fig. 1 should be shown, in addition to further information on how the material was prepared for the size assessment presented. Further details are also needed on the degree to which this preparation technique might have altered the dispersion and characteristics of the fibers; especially whether heating samples to such a high temperature of 480 degrees C might have altered their characteristics. The optical micrographs presented in Fig. 2 are of insufficient resolution to determine the physical nature of the MWCNT. At a minimum, we suggest toxicity studies using nanomaterials should include electron micrographs (TEM and SEM) of these materials that allow their shape, form and aggregation state to be assessed. These should ideally be augmented with additional physicochemical information on the material before and after administration. This is essential where small differences in nanotube construction, diameter, length, aggregation state and levels of impurities such as amorphous carbon and graphitic carbon, might lead to profound differences in toxicity. In summary, evaluation of the potential biological impact of nanomaterials such as MWCNT is essential, if they are to be developed and used safely. However, exposure doses must be realistic for the results to be relevant to issues of occupational or environmental health, and sample characterization needs to reflect the complexity and nanoscale structure of the material being studied. We believe the current study is flawed on both accounts.