This paper introduces a new active queue management scheme called Random Exponential Marking (REM), which aims to achieve high utilization and negligible loss and delay in a simple and scalable manner. The key idea is to decouple congestion measure from performance measure, such as loss, queue length, or delay. While congestion measure indicates excess demand for bandwidth and must track the number of users, performance measure should be stabilized around their targets independent of the number of users. REM has two key features: (1) match rate clear buffer, which attempts to match user rates to network capacity while clearing buffers (or stabilizing queues around a small target), regardless of the number of users; and (2) sum prices, where the end-to-end marking (or dropping) probability observed by a user depends on the sum of link prices (congestion measures) summed over all the routers in the path of the user. The first feature implies that high utilization is not achieved by keeping large backlogs in the network, but by feeding back the right information for users to set their rates. Simulation results demonstrate that REM can maintain high utilization with negligible loss or queuing delay as the number of users increases. The second feature is essential in a network where users typically go through multiple congested links. It clarifies the meaning of the congestion information embedded in the end-to-end marking (or dropping) probability observed by a user, and thus can be used to design its rate adaptation. REM differs from RED only in the first two design questions: it uses a different definition of congestion measure and a different marking probability function. These differences lead to the two key features mentioned in the last section. Simulation results show that REM performs better than DropTail and RED in wireline networks. REM also helps address the problem of poor TCP performance over wireless links, where TCP cannot differentiate between losses due to buffer overflow and those due to wireless effects such as fading, interference, and handoffs. REM can help by using ECN marking to convey congestion information to TCP sources. REM is implemented using local and aggregate information, and works with any work-conserving service discipline. It updates its price independent of other queues or routers, making it scalable and efficient. REM also has the advantage of being able to decouple congestion measure from performance measure, allowing for better performance and lower delay and loss. Simulation results show that REM can maintain high utilization and low delay and loss in equilibrium, even as the number of users increases. REM is also more effective than RED in wireless networks, where it can significantly improve the goodput of TCP. However, REM's transient behavior needs more careful study.This paper introduces a new active queue management scheme called Random Exponential Marking (REM), which aims to achieve high utilization and negligible loss and delay in a simple and scalable manner. The key idea is to decouple congestion measure from performance measure, such as loss, queue length, or delay. While congestion measure indicates excess demand for bandwidth and must track the number of users, performance measure should be stabilized around their targets independent of the number of users. REM has two key features: (1) match rate clear buffer, which attempts to match user rates to network capacity while clearing buffers (or stabilizing queues around a small target), regardless of the number of users; and (2) sum prices, where the end-to-end marking (or dropping) probability observed by a user depends on the sum of link prices (congestion measures) summed over all the routers in the path of the user. The first feature implies that high utilization is not achieved by keeping large backlogs in the network, but by feeding back the right information for users to set their rates. Simulation results demonstrate that REM can maintain high utilization with negligible loss or queuing delay as the number of users increases. The second feature is essential in a network where users typically go through multiple congested links. It clarifies the meaning of the congestion information embedded in the end-to-end marking (or dropping) probability observed by a user, and thus can be used to design its rate adaptation. REM differs from RED only in the first two design questions: it uses a different definition of congestion measure and a different marking probability function. These differences lead to the two key features mentioned in the last section. Simulation results show that REM performs better than DropTail and RED in wireline networks. REM also helps address the problem of poor TCP performance over wireless links, where TCP cannot differentiate between losses due to buffer overflow and those due to wireless effects such as fading, interference, and handoffs. REM can help by using ECN marking to convey congestion information to TCP sources. REM is implemented using local and aggregate information, and works with any work-conserving service discipline. It updates its price independent of other queues or routers, making it scalable and efficient. REM also has the advantage of being able to decouple congestion measure from performance measure, allowing for better performance and lower delay and loss. Simulation results show that REM can maintain high utilization and low delay and loss in equilibrium, even as the number of users increases. REM is also more effective than RED in wireless networks, where it can significantly improve the goodput of TCP. However, REM's transient behavior needs more careful study.