The authors discuss the process of asking and answering yes/no-questions in personal survey interviews from a conversational perspective. They examine the process with regard to yes/no-questions as given in the questionnaire and with regard to the yes/no-questions that interviewers may pose in the subsequent stages of a sequence, when they are probing on the respondent's initial answer or are trying to solve other problems. Hypotheses are derived from the co-operation principle and the politeness principle of conversation, and then empirically evaluated for survey interview settings. In the relatively informal stages of the answering process (subsequences) the conversations appear to go quite well according to the conversation rules, but in the beginning formal stage to a much lesser degree. In particular it has been observed that interviewers strongly prefer to ask one-sided positive yes/no-questions in the subsequences, and, secondly, that respondents usually give agreeing answers to these yes/no-questions. It is argued that this "normal" conversation strategy may seriously affect the validity of the information obtained, and, more generally, that the practical demands from the conversation rules on the interviewer's behaviour set limits on the researcher's abstract demands on that behaviour.