We present findings of a study carried out to determine the effect of streaming by gender on secondary school students' achievement in mathematics. In the study we analysed achievement scores on national examinations results for the years 1999 to 2001 of a sample of 1 489 candidates in four secondary schools in Nakuru District, Kenya. Raw data were analysed statistically and the hypotheses tested. Generally, the results indicated that streaming based on gender improved overall student achievement in mathematics and especially that of girls. Although further studies are needed to incorporate this result into official policy, there are strong indications that streaming by gender may be a useful class environment as an intervention towards improving the performance of girls in mathematics in co-educational schools.