The cancer stem cell debate over thelast 10 months has been an excitingtime for melanoma research. At theoutset of 2008, Markus Frank’s labora-tory published evidence that only atiny subpopulation of melanoma cells(approximately one cell in a million) hasthe capacity to maintain tumour pro-gression (Schatton et al., 2008). By thestandards of the day, it was an extre-mely important work, linking togethercancer stem cells, disease progressionand therapeutic resistance, and itseemed that a great leap forward hadbeen made in the fight against meta-static disease. However, at the closeof the year Sean Morrison’s groupraised the standards and presented evi-dence opposing the idea of a tiny sub-population of melanoma cells which arecritical for tumour progression, findinginstead that as much as a quarter of allmelanoma cells within any given lesionare tumorigenic (Quintana et al., 2008).The cancer stem cell (CSC) hypothe-sis rests on the premise that there area small number of specialized cancercells which combine the stem cell-likeproperties of limitless renewal andasymmetric division with the capacityto invade and populate regions distalfrom the original transformative event.The attractions of this hypothesisinclude that these cells may explainboth the cellular heterogeneity oftumours and resistance of metastaticdisease to current therapeuticapproaches. That melanoma metasta-ses are composed of different popula-tions of malignant cells is now wellaccepted; however, the explanation forthis heterogeneity has not yet beenestablished. Asymmetric division ofmelanoma stem cells to yield differentlycommitted progeny is a good model forpersistent heterogeneity throughoutdisease progression. Likewise, resis-tance to therapy may occur becausetumour cells are targeted via character-istics which, although widely expressedin melanoma tissue, may not beexpressed by a truly tumorigenic sub-population. The possibility that mela-noma stem cells underlie heterogeneityand therapeutic failure is plausible andthe seminal work by Schatton andco-workers appeared to bear this out.Challenges to the CSC hypothesis,include concerns about whether or notthe models used to evaluate CSCs arevalid. Xenograft models take advantageof compromised immune mechanismsto provide an essentially ‘black box’environment in which cancer cellproliferation may be experimentallymonitored. A typical CSC experimentseparates cells into fractions character-ized by surface markers known to beexpressed by normal stem cells, subcu-taneously injects them into immuno-compromised mice, and evaluates theircapacity to form tumours. It has beensuggested that xenografting underesti-mates the number of cells truly capableof driving tumourigenesis, because non-human environments may proliferativelyselect, in a non-informative fashion,some cancer cells over others. In theirstudy, Quintana et al. compared xeno-graft viability in the NOD⁄SCID mouseused by Schatton et al. with that of amore severely immunocompromisedmodel lacking natural-killer cell activity(NOD⁄SCID Il2rg