In this paper, a simple and eecient stop-and-wait (SW) automatic repeat-request (ARQ) scheme with adaptive error control is investigated. In this scheme, the channel state information (CSI) is extracted by monitoring the contiguous positive acknowledgment (ACK) or negative acknowledgment (NAK) messages. Exploiting this CSI, we adapt the coding strategy to the changes in the channel condition, and thus improve the throughput eeciency. In order to facilitate the throughput analysis and parameters optimization, we model the adaptive system by a Markov chain. Using this analytical model and assuming a static channel, an exact throughput expression for the adaptive ARQ protocol is derived and suboptimal adaptive system parameters are obtained. These design parameters are applied for the adaptive system in a typical time-varying mobile radio channel characterized by Rayleigh multipath fading on top of lognormal shadowing. The throughput performance of the proposed adaptive SW-ARQ scheme in such a time-varying fading channel is evaluated by computer simulation. For slow fading channels, the proposed adaptive system can track the channel variations very well, hence much throughput improvement can be achieved over conventional nonadaptive SW-ARQ schemes for almost all SNR values considered. The simulation results also connrm the applicability of the adaptive system parameters so-obtained by the throughput analysis in static channel, to a time-varying mobile radio channel.