A huge gap in science literacy is between students who do not show the competencies that are necessary to participate effectively in life situations related to science and technology and students who have the skills which would give them the potential to create new technology. The objective of this paper is to identify, for 25 countries, distinct subgroups of students with characteristics that appear to be associated with this proficiency gap. Data were based on the answers of 46,131 PISA 2006 students with scores classified below level 2 or above level 4, as well as the answers of their principals to school questionnaire and the OECD indicators of the financial and human resources invested in education at the national level for secondary school. The dependent variable of the analysis was a dichotomous variable the values of which represent the two different groups of students. The independent variables were the OECD indicators, and the items and indices derived from the student and school questionnaires. The analysis was based on classification trees and the findings were replicated and extended by the means of a multilevel logistic regression model. The results show that very specific levels of teachers' salaries, parental pressure on schools, school size, awareness of environmental issues, science self-efficacy and socio-economic status have a very important role in predicting whether 15 year olds in OECD countries will belong to the lower or the highest proficiency groups as regards their aptitude in the context of life situations involving problems of a scientific nature.