Treatment-free remission (TFR) has recently emerged as a goal of treatment in chronic myeloid leukaemia (CML). Molecular remission is sustained in around 30% of imatinib-treated patients who stop treatment after ≥2 years with undetectable minimal residual disease (UMRD) by conventional real-time reverse transcription polymerase chain reaction. An additional 20-30% of patients will lose UMRD, but remain in stable major molecular remission off treatment. Most patients with molecular recurrence have a significant increase in BCR-ABL1 within the first 6 months off treatment, but there are also rare late relapses. As re-treatment with imatinib restores control, a trial of TFR is safe so long as careful molecular monitoring is provided to enable prompt re-treatment. The minimum eligibility criteria for a trial of TFR are not yet defined, but the available data support a MRD level of around a molecular response of 4·5 log for at least 2 years. Factors associated with a higher probability of TFR include low risk Sokal score, prior interferon treatment, longer total duration of imatinib treatment and higher numbers of natural killer cells at the time of imatinib discontinuation. Preliminary data suggest that the rate of TFR in patients treated with more potent tyrosine kinase inhibitors will probably be higher. The biology that underlies TFR is an area of active investigation.