Systematic but unanticipated differences in patterns of responses to a test between two or more groups is generally taken as evidence of test bias. This study assessed whether a carefully-targeted instructional sequence could influence the effects of bias. A reading comprehension test with items previously identified as biased for certain groups was administered to two samples of minority children. This was followed a week later by two in-class sessions of the instructional intervention. Participants were then retested on the same instrument, first at the end of the same week in which they received the intervention, then again 4 weeks later. This pre/post/follow-up repeated-measures design allowed analysis, both statistical and graphic, of bias characteristics as they arose between groups at any given testing session and within groups across time. Results indicated that the test materials were generally very difficult. A few items improved significantly from pretest to posttest, although this improvement diminished somewhat across time. California Test of Basic Skill items which were targeted by the intervention showed a stronger degree of change for the treatment group than for the control group. (Author/PN)