[en] The apparatus and methods to measure PM short-time gain instabilities under the condition of pulsed light source simulating the operation of proton synchrotrons are described. Experimental results are presented for PM 84, PM 110, PM 115 and XP-2010. It is shown that short-time gain instabilities can be described by a simple mathematical model. It is found out that each type of PM is characterized by one or several constants of restoration time. It is convinient to characterize PM by the value of the maximum permissible stable current. For determination of this value the average anode current value may be chosen, when the long time (> 30 min) components of the gain instability do not yet appear. Following the measurement results it is established that I sub(stabl max.) <=1 μA for PM-85, 5μA - for PM-115, 10μA - for PM-84. It is found that for the majority of PM-85 the short-time instability is caused mainly by decrease in the effective quantum output, while for all the other studied PM the instability is connected with dinode ''fatigue''. The PM short-time instability can be lowered with amplifiers by taking off signals from dinode, placed sufficiently far away from anode