The glottal closure instant (GCI) detection is crucial in all kinds of speech processing applications. A variety of GCI detection methods were proposed but the performance of most existing methods was evaluated using only voiced parts of the speech signal. Under noisy conditions, the recorded speech consists of voiced, unvoiced and non-speech parts. Thus, most GCI detection methods demand an accurate voice activity detection (VAD) design. In this paper, we present post-processing techniques to improve the GCI detection accuracy and robustness in additive nonstationary noisy environments. We study the performance of the four methods: center of gravity (CoG), group delay function (GDF), zero frequency resonator (ZFR), and speech event detection using the residual excitation and a mean-based signal (SEDREAMS). The performance of each method with the proposed method-specific post-processing technique is tested and validated under both clean and noisy environments. Experimental results show that the method with post-processing technique outperforms the conventional GCI detection methods under noisy conditions. Results further show that the proposed technique can not only improve the overall detection accuracy but also reduce the complexity by avoiding VAD algorithm. © 2015 IEEE.