We describe the use of probit analysis to estimate breastfeeding indicators from current status epidemiological data. A health centre-based sample of 2411 children aged 0-1 year was investigated in Santo Andre, a large town in the Metropolitan Area of Sao Paulo, southeastern Brazil. Mothers were interviewed during routine pediatric consultations and asked about their current infant feeding practices. Probit regressions were calculated by a public-domain microcomputer programme written by one of us. The median duration of total (i.e. exclusive plus partial) breastfeeding in this children's sample, estimated as 108.8 days (95 per cent confidence interval: 95.5-123.2 days), is close to that recently reported in the city of Sao Paulo and nearby towns. However, the median duration of exclusive breastfeeding (28.9 days, 95 per cent CI: 17.9-38.3 days) is rather short when compared to recent estimates from this same region. Despite the nationwide efforts for promotion of exclusive breastfeeding, only 14 per cent (95 per cent CI: 12.3-17 per cent) of the infants were still being exclusively breastfed by 120 days of age. Therefore, a key feature of breastfeeding practices in this population sample, namely, the early introduction of supplementary foods, was identified by using simplified methods of data collection and analysis. This communication suggests that probit analysis of current status data may be further explored as a method for rapid epidemiologic assessment of breastfeeding practices.