Results obtained at Ithaca, N.Y., concerning radio reflections from the aurora by both pulse and continuous‐wave transmissions are described. At 104 Mc/sec, radar reflections show the following properties: (i) They occur only during auroras having ray structure; (ii) the radar must be directed roughly normal to the rays; (iii) echoes are complex, indicating contributions from many scattering elements mutually interfering in more or less random phase. At various frequencies, from 2.4 to 144 Mc/sec, measurements of fading show that auroral reflections fade at a rate roughly proportional to frequency; rates are between one and two powers of ten faster than would be expected for a quiet ionosphere.These observations are interpreted as indicating that radio reflections from auroras arise by scattering from numerous auroral columns of ionization, somewhat like meteor trails, having the magnetic zenith as radiant. The fading is principally due to a wind‐like motion of the auroral columns, much faster than is customary for meteor trails, and apparently much faster than can actually be observed for meteor trails occurring during auroral disturbances. Visual observations of the fine structure of rays with high‐power field‐glasses and by photometric analysis show that rays often consist of a large number of fine streaks, each of which extends only part of the length of the ray and lasts only about a second. It is tempting to associate these visual auroral streaks with the auroral ionization trails that are required to explain radio reflections from the aurora.Alternative interpretations suggested by other workers are considered, especially that by Harang and Landmark [11]. Reasons are given for rejecting these interpretations in favor of that described above.Although practically all of the auroral echoes observed at Ithaca are interpreted as coming fromE‐region levels, reasons are given why auroral echoes fromF‐region levels may be seen, especially by radars in lowe