Abstract In this paper, we perform modeling and analysis of the packet delay for a multichannel stop-and-wait automatic-repeat-request protocol (MSW-ARQ). In this protocol, the transmitter retransmits an erroneously received packet as well as all transmitted packets with larger sequence numbers than that associated with the erroneously received one. For this protocol, under the assumption that all parallel channels have the same transmission rate but possibly different error rates, we propose a method of analyzing the steady-state probability distribution function of the end-to-end delay, referred to as the packet delay, of an arbitrary packet. To demonstrate the generality of the proposed method, we further apply it to the packet delay analysis for a revised multichannel stop-and-wait ARQ protocol, referred to as MSW-ARQ-inS, in which only the erroneously received packets will be retransmitted. From expressions obtained through the analysis, numerical results are presented for chosen values of the number of channels and the error rates, based on which we conclude that MSW-ARQ-inS outperforms MSW-ARQ in terms of the mean packet delay performance. We also identify the performance trends of the mean packet delay with respect to system parameters. It is observed that the greater the variance in the error rates, the smaller the mean packet delay in both MSW-ARQ and MSW-ARQ-inS. Meanwhile, the number of parallel channels has only an insignificant impact on the mean packet delay, from which we conclude that the use of parallel channels is a favorable option to significantly increase the data transmission rate while keeping the mean packet delay at an acceptable level.