This study explored the differential performances of mathematics test items used to test secondary school girls and boys in the national examination. The main purpose was to find out whether type of items is the reason for girls' overachievement in the Malaysian mathematics national examination. To investigate seven types of items, Differential Item Functioning (DIF) analysis was used. DIF is a statistical method to identify items that function differentially for different groups of students of the same ability. Forty-six multiple choice items, categorised into seven types (Text, Graphics, Knowledge, Skills, Numbers, Shape and Space, Relationship) were obtained and certified by a panel of experts. The sample respondents were 1213 students aged around 16 to 17 years old. There were 529 boys and 684 girls who participated in this study. Data gathered were analysed using Winsteps version 3.48, a Rasch-based item analysis program. This study showed that gender-related DIF (GDIF) exists in the test used. GDIF in favour of girls was exhibited in 25 items, whereas 21 items were in favour of boys. Girls excelled in items categorised as Text and Numbers, whereas items categorised as Graphics, and Shape and Space were shown to be in favour of boys. Algebra seemed to be easier for the girls, while Statistics seemed to be easier for the boys. Results showed 14 items with significant GDIF, with sizes ranging from 0.29 to 0.65 (logits). Six items were significantly easier for the girls and eight for the boys. Although more items in the test were easier for girls, the study did not show evidence that girls' overachievement is due to imbalance usage of item type in the test. However, findings indicate that test developers should be sensitive of the occurrence of DIF and observe the proportion of item types showing DIF in all subjects tested in the national examination. (Contains 4 tables and 1 figure.)