This paper reports on an attempt to investigate Turkish primary school students' interest in science by using their self-generated questions. We investigated students' interest in science by analyzing 1704 self-generated science-related questions. Among them, 826 questions were submitted to a popular science magazine called Science and Children. Such a self-selected sample may represent a group of students who have a higher level of motivation to seek sources of information outside their formal education and have more access to resources than the students of low social classes. To overcome this problem, 739 students were asked to write a question that they wanted to learn from a scientist and as a result 878 questions were gathered. Those students were selected from 13 different schools at 9 cities in Turkey. These schools were selected to represent a mixture of socioeconomic areas and also to cover different students' profile. Students' questions were classified into two main categories: the field of interest and the cognitive level of the question. The results point to the popularity of biology, astrophysics, nature of scientific inquiry, technology and physics over other science areas, as well as indicating a difference in interest according to gender, grade level and the setting in which the questions were asked. However, our study suggests that only considering questions submitted to informal learning environments, such as popular science magazines or Ask-A-Scientist Internet sites has limitations and deficiencies. Other methodologies of data collection also need to be considered in designing teaching and school science curriculum to meet students' needs and interest. The findings from our study tend to challenge existing thinking from other studies. Our results show that self-generated questions asked in an informal and a formal setting have different patterns. Some aspects of students' self-generated questions and their implications for policy, science curriculum reform and teaching are discussed in this paper.