The statistical characteristics of amplitude masking for quantum noise stream cipher by intensity modulation is investigated, which is a Gaussian decomposition fitting based on the distribution probabilities of QNSC signals. The level amplitude distribution probability and standard deviation of noise of QNSC signal can be estimated, even if the encryption order is greater than 16, the decomposition of the probability density function is very complicated in calculation and produces considerable errors. Ultimate, an analysis of the performance of QNSC signal indicated that with the number of encryption order increasing, the error probability for Eve significantly increase, when the encryption orders are larger than 16, the error probabilities are higher than the hard decision forward error correction threshold 3.8 × 10−3. On the contrary, the error probability for Bob is always maintained at a relatively low level in this process. It is illustrated that analysis of amplitude statistical characteristics can estimate the security of the QNSC system effectively.