Magnetism in oxides was thought to be well-understood in terms of localized magnetic moments and double-exchange or superexchange rules. This understanding was shaken by the publication of an article in 2001 stating that thin films of anatase TiO2 with only 7 at.% Co substitution had a Curie point in excess of 400 K [Matsumoto et al., Science291, 854 (2001)]. Room-temperature ferromagnetism had previously been predicted for p-type ZnO with 5 at.% Mn [Dietl et al., Science287, 1019 (2000)]. A flood of reports of thin films and nanoparticles of new oxide “dilute magnetic semiconductors” (DMSs) followed, and high-temperature ferromagnetism has been reported for other systems with no 3dcations. The expectation that these materials would find applications in spintronics motivated research in this area. Unfortunately, the data are plagued by instability and a lack of reproducibility. In many cases, the ferromagnetism can be explained by uncontrolled secondary phases; it is absent in well-crystallized films and bulk material. However, it appears that some form of high-temperature ferromagnetism can result from defects present in the oxide films [Coey, Curr. Opin. Solid State Mater. Sci.10, 83 (2007); Chambers, Surf. Sci. Rep.61, 345 (2006)], although they are not DMSs as originally envisaged.