Using semi-structured interviews, I show how credit rating analysts themselves deal with this tension between officially declaring ratings as "opinions" and their highly sophisticated rating methodologies and practices. Two narratives can be extracted from the interviews conducted. Some analysts explicitly make a distinction between risk and uncertainty. They argue that there will always be a certain amount uncertainty that cannot be transformed into risk. Accordingly, their ideas to improve the rating process is to make it even more "qualitative" and less dependent on sophisticated financial mathematics. Others, especially analysts from smaller rating agencies suggest that it is exactly this qualitative aspect in the methodology of the big rating agencies that produces uncertainty, while their own – more stochastic and feedback sensitive – methods can calculate the risk of future default of financial products accurately.