Abstract A long-standing problem in meteor science has been the persistent presence of bias in the measured value of atmospheric scale heights obtained from radio meteor echoes. A common practice of fitting a linear function for bias correction follows the assumption that the systematic bias is devoid of seasonal asymmetry. This would be true if the mass and the velocity distribution of meteoroids remain invariant, both spatially and temporally. But so far no such convincing evidence has been published. On the contrary, fundamental arguments suggest that an universal mass function of radio meteoroids is counter-intuitive. This parameter cannot remain invariant due to the intrinsic variability in the meteor response function resulting from the Earth’s motion on the plane of ecliptic. In this paper, we show that an inverse relation exists between the width of meteor height distribution, expressed in unit of atmospheric scale height, and the exponent of the mass function. The overall mean of this exponent for the Sodankylä radar is 1.91 ± 0.02, modulated by a seasonal variation from the mean of the order of ∼±0.1. The stated inverse relation is applied to correct for the effect of mass distribution on the height distribution. Allowing for variable mass correction effectively removes the nonlinear bias in the measured scale heights in the meteor ionisation region.