Cancer-associated disturbances of regulated DNA methylation include both global hypomethylation and gene-specific (often even cancer-specific) hypermethylation. Both coexist and have become the subject of intense investigation. In haematological neoplasias, distinct sets of genes, including the p15/INK4b cell cycle inhibitor (mostly in myeloid malignancies) as well as p16/INK4a (only very infrequently in myeloid neoplasia), have been well characterised as to incidence of hypermethylation, concurrent gene inactivation and their re-expression following treatment with DNA methylation inhibitors. Several genes frequently methylated in haematological neoplasias have been studied with respect to their prognostic value. With the advance of low-dose schedules of demethylating agents (explored particularly in the elderly patient population) the rationale for reverting the 'hyper-methylator phenotype' has also prompted in vivo studies of gene reactivation following this type of treatment. However, ubiquitous surrogate markers for the efficacy of this type of treatment need to be developed. These may include reactivated haemoglobin F (HbF), as demethylating agents can result in clinically meaningful induction of HbF in patients with haemoglobinopathies. Because 'cancer testis antigens', which provide powerful signals for T cell cytotoxic activity on solid tumour cells, are usually silenced in leukaemia but can be reactivated in vitro and in vivo, they provide a rationale for an immuno-modulatory effect of demethylating therapy.