The sixth talk on machine translation is devoted to data, the fuel of statistical approaches to all NLP tasks. We discuss various possible data sources, effects of the text domain, the greed for more data and the diminishing utility of new data additions.