The distortion of the sample flux by a superconducting material produces a change of magnetometer output known as an image effect. Image effects for three Nb–Ti magnets were examined using a small current‐carrying coil as a source of a field‐independent and adjustable, controlled moment of a vibrating sample magnetometer. This allowed determination of the initial image effect (in zero field) and the field‐dependent image effect from zero to maximum magnetic field. The initial image effect (not detected by conventional calibration) can be large, and the field‐dependent part was 0.25–0.35 of initial image effect. The variation of the field‐dependent image effect on field sweep rate and magnet temperature, and the change of the spatial distribution of the detection coil sensitivity were also observed. The image effect is independent of sample moment and decreases rapidly with decreasing detection coil radius. The method of images was employed to calculate the image effect for perfect shielding as the function of detection coil dimensions relative to the magnet bore. For a small superconducting volume fraction &lgr;, the initial image effect for multifilamentary wire magnets is smaller than calculated by an approximate factor &lgr;/(1−&lgr;) in agreement with observations for the three Nb–Ti magnets. Some influences of the multifilamentary structure of the superconducting wire on the image effect are discussed.