The purpose of this study was twofold: to investigate to what extent characteristics of anchor tests may affect precision of item calibration, and to estimate to what extent precision of item calibration may be affected by removal of persons whose response patterns deviate from those normally expected from the Rasch one-parameter logistic model. Three characteristics of anchor tests were under consideration: the number of anchor items, and the range and average of difficulties of the anchor items. The data were taken from the nation-wide norming data of the Otis-Lennon School Ability Test, Form R, Intermediate Level. Anchor test characteristics did not show systematic effects on final calibration results. The removal of misfitting persons was detrimental to calibration results. Further studies are needed to clarify the effects of anchor test designs and person fit on linkings. (BW)