We performed an epidemiological study on 1,471 ankylosing spondylitis patients treated with repeated intravenous injections of the short lived alpha-emitter (224)Ra (excluding radiation therapy with X-rays) between 1948 and 1975. These patients have been followed together with a control group of 1,324 ankylosing spondylitis patients treated neither with radioactive drugs nor with X-rays. The mean follow-up time was 26.3 years in the exposed and 24.6 years in the control group. To date, causes of death have been ascertained for 1,006 exposed patients and 1,072 controls. Special emphasis was placed on the reporting of malignant diseases. Expected numbers of cases were computed for the age, sex and calendar year distribution of both groups using cancer registry incidence rates. In the exposed group 18 cases of kidney cancer (vs. 9.1 cases expected, P < 0.01) and 4 malignant thyroid tumours (vs. 1.2 cases expected, P = 0.03) were observed. In the control group the observed cases for these tumours were not significantly elevated. The most striking observation, however, were the 21 cases of leukaemia in the exposed group (vs. 6.8 cases expected, P < 0.001) compared to 12 cases of leukaemia in the control group (vs. 7.5 cases expected). Further sub-classification of the leukaemias demonstrated a high increase of myeloid leukaemia in the exposed group (12 cases observed vs. 2.9 cases expected, P < 0.001), and out of these, especially a high excess of acute myeloid leukaemias (7 cases observed vs. 1.8 expected, P = 0.003). In the controls the observed cases are within the expected range (4 myeloid leukaemias vs. 3.1 cases). This increase in total leukaemias as well as particularly in myeloid leukaemias is significant in direct comparison between the exposed and control groups too (P < 0.05). The enhanced leukaemia incidence in the exposed group is in line with the observation of increased leukaemia incidence in mice injected with (224)Ra.