In a four-cavity velocity modulation tube comprising an electron gun assembly, an input cavity, a pre-intermediate cavity, a post-intermediate cavity, an output cavity, a collector assembly and drift tubes interposed between adjacent ones of said cavities, said input cavity is tuned to the proximity of the lower end of the operating passband of said velocity modulation tube, said pre-intermediate cavity is tuned to the proximity of the upper end frequency of said operating passband, said post-intermediate cavity is tuned to a fundamental resonant frequency that is higher than the upper end frequency of said operating passband but lower than a frequency which is 1.6 times as high as the center frequency of said operating passband, the normalized length of all of said drift tubes being smaller than 90 degrees in terms of the reduced plasma angle, and the normalized length of the drift tube positioned between said pre-intermediate cavity and said post-intermediate cavity being larger than the normalized lengths of the other drift tubes.