Gender differences in most cognitive skills are fading, but a gender g
ap remains in secondary school that favors males in higher level math
skills. This gap is not evident in elementary school where test scores
for the two sexes are equivalent. However, the daily experiences of y
oung boys and girls differ in ways that could affect their math skills
in early adolescence. In a large random sample of youngsters in Balti
more, over their first two years of school, boys' gains in math reason
ing achievement were more sensitive to resources outside the home than
were girls'. In line with the greater responsiveness of boys' math sk
ills to these neighborhood resources, the boys' math reasoning scores
became significantly more variable over time than did the girls'. When
differentiated course programs became available in middle school, thi
s greater variability of the boys' math scores led the high-scoring bo
ys in the ''academic'' program to outscore the girls in that program,
even though in the total sample the means for boys and girls were abou
t the same. In short, by the end of middle school a ''gender gap'' eme
rged in math among high-scoring youngsters. These trends in variabilit
y and the greater sensitivity of males to neighborhood resources combi
ned with school tracking offer a new and more sociological perspective
on the emergence of the gender gap in math in early adolescence.