This study examined irregularities in reading comprehension (RC) test results at both the individual and aggregate levels, focusing on the effect of test speededness on test scores. Data were taken from results of the Stanford Achievement Test--Seventh Edition (SAT/7), administered during April of 1983 to about 50,000 Palm Beach County (Florida) students in grades 2 through 9. This study concentrated on fourth graders. To examine the effect of speededness, the performance on other SAT/7 subtests of students who completed the RC subtest was compared to that of students who did not finish the subtest. Individual student and classroom score profiles were grouped on the basis of RC subtest completion rates. Groups that completed the RC subtest scored lower on other battery subtests (e.g., language) than did those that did not complete the RC measure. Given these results, the test publisher agreed to generate norms based on the first 40 items of the 60-item RC subtest, using the Rasch model. Rescoring the RC subtest with the 40-item norms substantially corrected score irregularities. Results suggest that test speededness is one aspect of measurement that can affect test meaning and interpretation materially. Four figures and four data tables are included. (Author/TJH)