Item responses by adult Egyptian native speakers of Arabic on the ALIGU test battery were analyzed statistically to determine which items deviated significantly from the expected frequency of correct responses, and which item distractors were chosen most frequently. Thirty-two of 100 grammar responses showed significant incorrect responses, and on each, students chose a particular distractor more often than could be expected by chance, initially suggesting native language interference. Of these 32 items, only 8 occurred frequently in the lowest proficiency group, and 7 occurred only in the highest proficiency group. Seventeen appeared only in the middle proficiency group's responses. Clear evidence of Arabic interference appeared in 23 of the 32 items. Another four showed some possible interference, and five showed none. Of those five, two or three were among the items with which even the most proficient speakers had difficulty, and all showed errors characterized as developmental and not interference-related. The results suggest that the less proficient subjects relied more on their native language than the more proficient subjects. The test validates earlier research, and identifies certain grammatical structures that should be taught with a "remedial approach" using review, contrast, and re-review techniques. (MSE)