This study examined the effects of motion of an artificial bed partner on the sleep of a real subject on the same mattress. A 240-lb eight-sided cylindrical roller served as the artificial bed partner. The roller was placed on one side of a king-size mattress and a normal adult slept on the other half. On experimental nights, the roller intermittently rocked back and forth for brief periods of time which in aggregate represented approximately 13% of total sleep time. On control nights, the roller was stationary. After an adaptation night in the laboratory, subjects spent two more nights in the sleep laboratory. The order of Roller On and Roller Off nights was random. 24 subjects were randomly assigned to sleep on the three surfaces. The three mattresses used in the study had different construction designs and in turn variable properties of motion transfer. Sleep variables were measured objectively by nocturnal polysomnography and subjectively by a postsleep questionnaire. Analysis of the data indicated that motion transferred laterally across a mattress was associated with a significant increase in Stage 1 sleep, and a significant decrease in Stage 3/4 sleep. Over-all, sleep efficiency, number of awakenings, and wakefulness after sleep onset did not show significant changes from control to experimental nights. The amount of change in sleep architecture, however, was more notable on the mattresses which transferred more motion. An implication of these findings is that motion transferred across the surface of a mattress, by lightening the depth of the sleep and thereby decreasing the auditory arousal threshold, can increase the potential for sleep disruption from environmental stimuli such as noise.