In Slovenia we have centralized school system with approx. 450 public elementary schools with grades from 1 to 9. Only one elementary school is private, using Steiner's methods of education (Waldorf's school). There are three universities for studying teaching, but majority of teachers come from Ljubljana University since faculties for education on other two universities were established in recent years. Curriculum is the same for all elementary schools. Children usually go to school which is near home. Because all of the above we would expect that knowledge achievement is more or less the same across the country. According to the PIRLS 2001 data this is far from reality. Average reading achievement in Slovenia was at the international average of 500 points. Some schools did not go beyond 430 points (which means that average child at the 3rd grade at this school is illiterate) while some schools achieved good reading achievement of 570 points. From PIRLS 2001, TIMSS and many other surveys we know that better SES means higher achievement. However, this is true in general. PIRLS 2001 data show that there are a lot of exceptions to this rule. We chose 10 "good" and 10 "bad" schools according to PIRLS 2001 reading achievement and socio-economic status of students - SES. SES was described by parameters of family income, parents' education, self perception of parents etc. If school achievement had been much higher or much lower than expected according to SES of students, a school was picked up. At the moment we are interviewing (structured interviews) and collecting questionnaires from teachers, librarians and school masters of the chosen schools. We hope to find some pattern which separate good and bad schools, if such pattern exist at all. Results would be presented at the ECER conference