This study investigates the relationship between an examinee's test anxiety level, degree of person fit, and ability. Three major types of person fit indices measure the degree of unusual response patterns: norm comparison indices; goodness-of-fit indices; and extended indices. Five different person fit indices were calculated for the results of the Metropolitan Achievement Test (MAT) reading, mathematics, and science subscales of 413 seventh and eighth grade students: (1) the modified caution index; (2) the personal biserial correlation; (3) the norm-conformity index; (4) the Rasch person fit statistic; and (5) the extended caution index. The Test Anxiety Scale for Adolescents (TASA) was administered approximately two weeks prior to the MAT. Data were analyzed by a linear multiple regression analysis. High intercorrelations among the person fit indices confirmed previous finding. Significant interactions between ability and test anxiety demonstrated the moderating effect of ability levels on the person fit measures and test anxiety relationship. There was a direct relationship between person fit measures and test anxiety for lower ability examinees and an increasing negative relationship for high ability examinees. For examinees in the average ability ranges, there is little or no relationship between test anxiety and person misfit. The results are interpreted using the Cognitive-Attentional Theory of Test Anxiety. (BS)