The Barkhausen noise was measured in nonoriented Fe-3%Si steel with different average grain sizes. Air gaps between the yoke and the measured objects were also varied during the measurements. The change of the grain size was achieved by different combinations of cold rolling and heat treatment processes. The rise of the gap size degraded the level of the Barkhausen noise and caused that the parameters of the Barkhausen noise, such as the amplitude of the Barkhausen noise's envelope, decreased. In order to suppress the influence of the air gap size on the measurement results, we analyzed the amplitude probability distribution of the Barkhausen noise and we found that the distribution at small levels of the voltage practically does not change with the grain size, but it increases with rise of the gap size. This change of the amplitude probability distribution with the gap size was used to correct the amplitude of the Barkhausen noise's envelope. It was shown that the correction essentially increases the precision of the evaluation of the grain size at varying air gap. Similar technique can be used also to decrease the error of evaluation of other microstructural changes of ferromagnetic materials using the Barkhausen noise method, without measuring and feedback setting the defined waveform of the magnetic field in the sample. For this method to be applicable, several conditions should be met, especially the level of the Barkhausen noise should be essentially higher than the level of the disturbing noise.