Carcinoma in situ (CIS) of the breast has increased many-fold in incidence rates and as a proportion of new breast cancers following the introduction of mammographic breast screening. To provide population-based estimates of invasive breast cancer risk following CIS, we linked data on 249 incident primary CIS (median age 53 years) to the Cancer Registry of the Swiss Canton of Vaud (about 600,000 inhabitants) over the period 1977-1994. Women with concurrent invasive cancers of the breast were not included. Standardized incidence ratios (SIR) were determined according to the exact Poisson distribution, with stratification for age and year of diagnosis. A total of 24 cases of breast cancer vs. 3.4 expected [SIR = 7.2, 95% confidence interval (CI): 4.6-10.6], and 7 cases of other neoplasms (except non-melanomatous skin cancer) vs. 6.9 expected (SIR=1.0, 95% CI: 0.4-2.1) were observed. The SIR was 10.4 during the first year, 5.6 between I and 4 years, and 7.7 after > or = 5 years after CIS diagnosis. SIRs were consistent in women below and above age 55 years, but somewhat higher for ductal (SIR=8.6) than lobular (SIR = 4.2) CIS. Six deaths from breast cancer were observed vs. 1.5 expected (standardized mortality ratio=4.0, 95% CI: 1.5-8.7). In 13/19 ductal CIS, but in 2/4 lobular CIS, invasive cancer occurred in the same breast. In most women, CIS and subsequent invasive cancer showed the same morphological (i.e., ductal or lobular) features. The cumulative risk of breast cancer was 16% 10 years after CIS diagnosis, emphasizing the importance of adequate surveillance of women after CIS of the breast.