In Malaysia, while the achievement gap between rural and urban schools is flattening over time, the gender achievement gap is widening. This study therefore re-examines the phenomenon of "Lost Boys" using data from the Trends in International Mathematics and Science Study (TIMSS). Gender differences in disruptive behaviour and disciplinary challenges may widen boy-girl academic performance difference and, if so, differential exposure to crowded classrooms can play a role. Alongside other hypothesis, therefore, we specifically explore the gender differentiated effects of class sizes on student achievement. Methodologically, we employ the two-step least squares and quantile regression methods. Findings show that decrease in class size statistically does not have a positive effect on student scores in mathematics and science regardless of gender. If anything, there is evidence that only high-achieving female students particularly from urban schools and advantaged socioeconomic backgrounds may benefit from reduction in class size incentive. While improving the quality of current teachers and certain school infrastructure such as school buildings, lightning systems, and classrooms as well as granting teacher incentives appears to improve boys' achievement scores, on the balance, it is unlikely that the boy-girl difference in maths and science test scores in Malaysia can be primarily explained by school level factors. We conclude by discussing alternative explanations related to our findings on class size as well as other cost-effective policy responses and non-school factors to tackling the "Lost Boys" phenomenon.