Experimental measurements of threshold contrast (Csub(T)) as a function of air kerma rate at the input plane of the image intensifier have been made for several diagnostic fluoroscopy units in clinical use. Threshold contrasts are determined by viewing a test object containing holes of fixed diameter and various depths under defined irradiation conditions. Kerma rate variations are effected by introducing aluminum sheets into the x-ray beam at fixed values of tube potential and current. At low kerma rates where quantum noise dominates, low tube potentials (60 kVsub(p)) usually yield lower values of Csub(T) than do higher potentials (100 kVsub(p)). At higher kerma rates the opposite is often true. A simple theoretical model for noise propagation in fluoroscopic imaging systems using models of diagnostic x-ray spectra lends qualitative support to the experimental findings. The often-quoted suggested upper limit of 100 ..mu..R s/sup -1/ (0.87 ..mu..Gy s/sup -1/) at the input phosphor would seem to be justified under the test conditions since little improvement in Csub(T) is usually observed at higher kerma rates. However, application to clinical practice would ideally require the use of more realistic phantom studies.