The effects of photomultiplier voltage, discriminator setting, and temperature on the signal-to-noise ratio in a photon counting experiment have been measured for several photomultiplier types. The photomultipliers tested were an RCA 1P28A, an RCA 4832, an RCA 1P21, and a Hamamatsu R406.The results Indicate that photon counting experiments should be performed at the hlghest possible photomultlpller voltage. The data also show that the increase in the signal-to-ndse ratio realized when the photomultiplier is cooled is very dependent on the photomultiplier type.