Gaps in educational attainment have proven to be important as they affect other outcomes of interest such as college education choice, professional careers and labor market outcomes. In this paper I analyze how household composition affects the gender gap in standarized test score, focusing on being an only child or a first born in a multiple children family. I use 2003 data from a mandatory exit exam of secondary education for the entire universe of Colombian students. My results show that boys outperform girls in mathematics, specially in the upper part of the score distribution, while girls outperform boys in language but in the bottom part of the score distribution. Interestingly, only child girls do better than girls with siblings: both by closing the gap in mathematics in the upper part of the score distribution and widening it at the bottom part of the distribution in language. On the other hand, being a first born girl has a negative effect for high performing students in math.