Monte Carlo method recently has been in the limelight as a very useful method to solve problems of operations research and other fields. In order to effectively apply Monte Carlo method, it is indispensable to generate high speed random numbers which can be easily generated from binary random numbers. One of the practical and very useful methods to get them is to use as the random numbers' sources the physical fluctuation phenomena such as noise. In this paper, using the binary random numbers' checker, the statistical properties are investigated on the high speed binary random numbers' trains. The radom pulse train, which is used to generate binary random numbers and which has long been believed to follow the poisson distribution, is proved, if strictly speaking, to follow other distribution than the poisson distribution. So, new distribution is assumed for the random pulse train and experimental and theoretical values are discussed under this new assumption.