This study was designed to determine item bias by sex on a test of spatial visualization administered to middle school students. Two chi square techniques were used to assess item bias: the Scheuneman C2 and the Camilli chi square. The direction of bias was indicated by assigning positive signs to the chi square components contributed by one group and negative signs to the other. A computer program to carry out the chi square bias detection techniques was written in FORTRAN 5 to run on a Cyber 750 computer. The program permitted the user to vary the total number of subjects, items, and status groups. The Camilli method usually identified more items as biased than did the Scheuneman method. It tended to place items at a lower probability level. The chi square item bias techniques were shown to have potential for small scale test users and were particularly suitable for mastery and criterion-referenced tests. (DWH)