When compared to satellite detector measurements of dust particles of mass < 10^-6 g and optical meteor observations for mass > 10^-2 g, the flux of the interstitial radar meteors is discrepant: the radars render fluxes which seem too small by a factor of about 20-30. This has usually been explained as being due to the majority of the flux being held in low-velocity meteors which produce little ionization and hence have limited radar detectabilities. We propose an alternative hypothesis: that the discrepancy is due to wavelength-dependent effects, implying that conventional meteor radars {f > 20 MHz) only detect the lower-altitude underdense meteors. To test this we have determined the height distribution of radar meteors at 2 and 6 MHz, at which frequencies the echo ceilings are much higher than the 100-105 km limits of VHP radars. We find that the distributions peak at “105 km, fully 10 km above the peaks of VHP radars, with many meteors occurring to at least 140 km altitude. Additional observations using the powerful Jindalee radar in central Australia confirm these results, and show that the cumulative flux of particles of mass > 10"-6 g is about 9 x 10^-8 m^-2 sec^-1; this is consistent with satellite data and is over an order of magnitude larger than derived in previous radar meteor experiments.