For the treatment of differentiated thyroid cancer, surgery, radioiodide therapy, and thyrotropin-suppressive thyroxine application represent established therapeutic measures of proven efficiency, affording a good prognosis for this disease. However, in up to 30% of the cases, dedifferentiation is observed, giving rise to tumors that are refractory to conventional treatment. Eventually, this may lead to the most malignant human tumor, anaplastic thyroid carcinoma, with a life expectancy of only a few months after diagnosis. Among novel approaches for the treatment of dedifferentiated thyroid carcinomas, retinoic acid redifferentiation therapy was evaluated in several in vitro and in vivo studies. Cell culture experiments in thyroid carcinoma lines show that RA treatment affects thyroid specific functions (type I 5'-deiodinase, sodium/iodide-symporter), cell-cell or cell-matrix interaction (intercellular adhesion molecule-1, E-cadherin), differentiation markers (alkaline phosphatase, CD97), growth, and tumorigenicity. The observed changes, which involve multiple parameters that characterize a mature, functional thyrocyte, may be interpreted as partial redifferentiation. In clinical pilot studies, about 40% of the patients responded to RA application with an increased radioiodide uptake. In an evaluation of 20 RA-treated patients with well-documented data sets, 8 exhibited a decrease (4) or stabilization (4) in tumor size and/or in serum thyroglobulin levels in addition to enhanced radioiodide transport. This indicates that these patients with a long history of unresponsiveness to other treatment may have experienced an actual therapeutic benefit. These data suggest that RA redifferentiation therapy, considering especially its comparatively mild side effects, may soon represent an alternative therapeutic approach to otherwise untreatable thyroid tumors.