Covert timing channels, as a growing threat of network security, provide the possibility of leaking confidential information to an attacker. Thus, detection and countering these channels are important as a defensive measure in computer networks. The attacker uses an appropriate encoding schema to modulate covert information on temporal features of network packet stream. Embedding covert information in net- work traffic, casuses changes in traffic statistical properties such as distribution, correlation and entropy, which can be used in detection of covert timing channels. In this paper, statistical methods are identified and analyzed to detect these channels, and two encoding schemas L-bit to N-packet and non-detectable are used to implementi covert timing channels and those are evaluated. The results show that by using Kolmogorov-Smirnov, Regularity, corrected Entropy and corrected Conditional Entropy tests, we are able to completely detect L-bit to N-packet channel, and the stealthiness of non-detectable timing channel can be proved in a practical evaluation as well.