Knowledge of the molecular events that govern human thyroid tumorigenesis has grown considerably in the past ten years. Key genetic alterations and new oncogenic pathways have been identified. Molecular genetic aberrations in thyroid carcinomas bear noteworthy resemblance to those in acute myelogenous leukemias. Thyroid carcinomas and myeloid leukemias both possess transcription factor gene rearrangements-PPARgamma-related translocations in thyroid carcinoma and RARalpha-related and CBF-related translocations (amongst others) in myeloid leukemia. PPARgamma and RARalpha are closely related members ofthe same nuclear receptor subfamily, and the PML-RARalpha and PAX8-PPARgamma fusion proteins both function as dominant negative inhibitors of their wild-type parent proteins. Thyroid carcinomas and myeloid leukemias also both harbor NRAS mutations (15-25% of both cancers) and receptor tyrosine kinase mutations--RET mutations in thyroid carcinomas and FLT3 mutations in myeloid leukemias. The NRAS and tyrosine receptor kinase mutations are not observed in the same thyroid carcinoma or leukemia patients, suggesting that multiple initiating pathways exist in both. Lastly, thyroid carcinomas and myeloid leukemias possess p53 mutations at relatively low frequency (10-15%) in patients who tend to be older and have more aggressive, therapy resistant disease. Such parallels are unlikely to occur by chance alone and argue that common mechanisms underlie these diverse epithelial and hematologic cancers. The comparison of thyroid carcinomas and myeloid leukemias may highlight areas of thyroid cancer investigation worthy of further focus. For example, few collaborating mutations have been defined in thyroid carcinomas even though they play a clear role in myeloid leukemias, as exemplified by RARalpha rearrangements and FLT3 mutations that together dictate the promyleocytic leukemia phenotype. Functional interactions between collaborating mutations are possible at multiple levels, and it is tempting to speculate that some thyroid carcinomas might develop through an unique combination or co-activation of RET and RAS and/or RET and PPARgamma (and/or other) signaling systems. In fact, the ELE1-RET (PTC3) fusion protein contains the ELE1 nuclear receptor co-activator domain and it appears to physically associate with and inhibit wild-type PPARgamma in some papillary carcinomas. The similarities of the fusion proteins in thyroid carcinoma and myeloid leukemia suggest that a more directed search for fusion genes in non-thyroid carcinomas is warranted. In fact, novel fusion genes have been identified recently in aggressive midline, secretory breast, and renal cell carcinomas, although the epithelial nature of the latter is not well-documented. Interestingly, these cancers all tend to present more frequently in adolescence and young adulthood in a manner similar to thyroid and myeloid malignancies that have fusion genes. The analyses of cancers that present earlier in life may enhance fusion gene recognition in other carcinoma types. Definition and biologic characterization of the precursor cells that give rise to thyroid carcinoma will also be important. Myeloid leukemias are thought to arise from stem/progenitor cells that acquire disturbed self-renewal and differentiation capacities but retain characteristics of the myeloid lineages. Although the presence of comparable stem/progenitor cells in the thyroid are not defined, distinct thyroid cancer lineages and patterns of differentiation exist and candidate stem/progenitor cells such as the p63-immunoreactive solid cell nests are apparent. A last important area is development of molecular-based therapies for thyroid carcinoma patients resistant to standard radio-iodine treatment. Treatments for such cancers are limited and pathways defined by thyroid cancer mutations are prime targets for pharmacologic interventions with molecular inhibitors. Tyrosine kinase inhibitors and nuclear receptor ligands have proven dramatically effective in some myeloid leukemia patients. Various molecular inhibitors are being investigated now in thyroid cancer models. Such developments predict that the thyroid cancer model will continue to provide biologic insights into human carcinoma biology and that improved pathologic diagnosis and treatment for thyroid cancer patients sit on the not too distant horizon.