In a register study all women in the West of Sweden Health Care Region with a breast cancer diagnosed between 1960 and 1980 (n = 13,490) were followed up in the Swedish Cancer Register to the end of 1988 for later occurrence of a soft tissue sarcoma (STS). Nineteen sarcomas were reported, whereas 8.7 were expected and the relative risk (RR) was 2.2 (CI 95% 1.3-3.4). The absolute risk was 1.7/10(4) person years (PY) in comparison with 0.8 expected. To obtain a more detailed analysis of the associations between arm lymphoedema, radiotherapy and STS development, and to control the quality of the register data, a case control study was also performed. Clinical records from the different hospitals in the region were collected for all the 19 cases as well as for three selected controls per case. The histopathology of the cases were reviewed, and one of the cases was reclassified as a malignant melanoma and excluded from further analysis. Thirteen of the cases were clustered around the treated breast area. To quantify the exposure to radiotherapy, the integral dose was estimated. The presence of lymphedema was included as a binary variable in the analysis. The exact conditional randomisation test indicated a significant correlation between the integral dose and the development of an STS (p = 0.008) and this association was still significant after stratification for arm oedema. A conditional logistic regression analysis with STS as the dependent variable and the integral dose as the explanatory variable gave an odds ratio (OR) of 5.2/100 J (CI 95% 1.3-21.2), and if this regression was restricted only to the STS developing in the radiation fields the OR was 3.2/100 J (CI 95% 0.8-12.9). Thus, the excess of STS in this breast cancer cohort was very low (0.9/10(4) PY). However the integral dose correlates well to the development of STS and can be useful in quantifying even small risks of secondary malignancies in the breast cancer population.