Among 41 participating countries in the Programme for International Student Assessment (PISA) 2003, the gender difference in favour of females was greatest in Iceland in the three subjects tested: mathematics, science and reading. The aims of this article are to put these findings in national and international context, and report on a number of attempts to explain them. This large gender difference is confirmed and given concurrent validity through the results of the annual Icelandic national standard tests in these subjects. The female advantage is apparent at 4th and 7th Grade, depending on subjects, and is maintained through university level. A comparison of PISA 2003 results with PISA 2000 and recently with PISA 2006 indicates that the year 2003 was somewhat exceptional. However, the basic stability of the gender differences over the past years is demonstrated. A number of explanations for the overall female superiority in PISA and the national standard test performance in Iceland are examined, namely gender difference on low versus high stakes tests, regional effects, school variability and psychological factors. No evidence is found for a stable school effect on gender differences across years, stable regional differences or for the explanation that male disadvantage only appears in "low stakes" exams like PISA. Various psychological factors such as anxiety and self-esteem are shown to have stronger links with academic performance for girls compared with boys.