The main objective of the study was the development of a semi-standardized test in Mathematics at grade 9 through Rasch modelling. The population of the study was 59,168 students of grade 9 admitted to 718 secondary schools of Bahawalpur Division (Pakistan) during session 2010-11. The schools were categorised gender-wise, locality-wise and ownership-wise to ensure representation of the population in the sample. Cluster sampling was used to identify the sample. For this, 3 schools were randomly opted from each category and all the students of grade 9 in these schools were included in the sample. There were 720 students. However, 642 students participated in the study due to different reasons. Two equivalent test forms, each with 43 items were developed from Mathematics of grade 9. These were ultimately transformed into a 86-items test to be semi-standardized. The content validity of the test was obtained by incorporating the opinion of teachers in Mathematics and researchers. The alternate-forms reliability of the test was 0.92. The two test forms were administered everywhere with uniformity of instructions and arrangements for two consecutive days to retain the external validity of the test. The Rasch analysis showed more accuracy thoroughly in the appraisal of test items than respondents in terms of logit values, Model error estimates and fit-statistics. The major recommendations of the study were the replication of the current study, its execution at other times during the academic session, and finding probability correct for all items and persons to see response patterns more realistically.