Sleep and ongoing cycling of sleep states are required for neurosensory processing, learning, and brain plasticity. Many aspects of neonatal intensive care environments such as handling for routine and invasive procedures, bright lighting, and noise can create stress, disrupt behavior, and interfere with sleep in prematurely born infants. The study empirically investigated whether a 30-minute observation of infant sleep states and behavior could differentiate an intervention to promote sleep in premature infants with feeding difficulties relative to conventional care (standard positioning, standard crib mattress [SP]). We included an intervention to determine the ability of the method to discriminate treatments and generate a benchmark for future improvements. The intervention, a conformational positioner (CP), is contoured around the infant to provide customized containment and boundaries. To more fully verify the 30-minute observational sleep results, standard polysomnography was conducted simultaneously and sleep outcomes for the 2 modalities were compared. In a randomized crossover clinical trial, 25 infants, 31.5 ± 0.6 weeks' gestational age and 38.4 ± 0.6 weeks at the study, with gastrointestinal conditions or general feeding difficulties used each intervention during an overnight neonatal intensive care unit sleep study. Infant sleep states and behaviors were observed during two 30-minute periods--that is, on the positioner and mattress--using the naturalistic observation of newborn behavior. Two certified developmental care nurses assessed sleep state, self-regulatory, and stress behaviors during 2-minute intervals and summed over 30 minutes. Sleep characteristics from standard polysomnography were measured at the time of behavior observations. Infants on CP spent significantly less time in alert, active awake, or crying states by observation compared with SP. Surgical subjects spent more time awake, active awake, or crying and displayed a higher number of behavior state changes than the nonsurgical infants. The percentage of time in observed deep sleep and quiet sleep was correlated with both percentage sleep efficiency (r = 0.78) and fewer state shifts per hour (r = -0.65) from electroencephalogram (EEG). Sleep efficiency by EEG was greater on CP versus SP. The CP enabled sleep compared with the standard mattress (SP) over 30-minute observation periods. Sleep status from behavioral observation was verified by standard EEG-based sleep techniques. Behavioral observation of sleep states may be a useful strategy for measuring the effectiveness of strategies to facilitate sleep in premature infants. Surgical subjects may benefit from additional interventions to promote sleep.