Non-invasive blood pressure measurements by the auscultatory method do not provide reliable, reproducible blood pressure values in many neonates because the Korottkoff sounds are often very difficult to detect. This resulted in the development of many alternative indirect methods. Devices utilizing the Doppler ultrasound technique have not found wide acceptance. Since the introduction of automated oscillometric blood pressure monitors, arterial blood pressure has been increasingly brought into discussion as an indicator of the circulatory state. This is the first study to investigate the similarity and reproducibility of the data obtained with five oscillometric devices for measurement of blood pressure in neonates. MATERIALS AND METHODS. Since investigations on the technical performance are not practical in the clinical setting of a neonatal or pediatric ward, we used two simulators, the CuffLink (Dynatech, Nevada, USA) and a device developed by the PTB (Physikalisch-Technische Bundesanstalt). While the latter uses oscillations originally obtained from neonates, the CuffLink uses artificial and therefore ideal signals for the blood pressure monitors. The signals used for the PTB simulator were obtained from three neonates with an average age of 2 months and a weight of 3.5 kg, 4.4 kg, and 7.8 kg. The following blood pressure monitors were studied: Hoyer/Colin, BP-1001; Datex, Cardiocap II; SpaceLabs, model no. 90426; Hewlett-Packard, NBP M1008A; Critikon, Dinamap 1846. Before the measurements were started, the cuff pressure display of each monitor was checked according to a verification procedure. Although the 4 mmHg margin of error was not exceeded, the results were corrected. RESULTS. The results of the measurements show significant differences between the blood pressure monitors from the various manufacturers, with the differences for the ideal signals of the CuffLink-Simulator being less pronounced than those for the PTB simulator. Direct comparison of results is therefore often impossible. The standard deviation, taken from 20 measurements per monitor and simulation, is below 4 mmHg for both simulators. We can therefore conclude that the reproducibility of data is satisfactory and the emerging trend is reliable. DISCUSSION. The difference between the results of the PTB simulator and the CuffLink are probably due to the method of evaluation and the identification of artifacts of each blood pressure monitor. This is also confirmed by the studies of Mieke et al. The manufacturers should provide devices that display comparative results and improve the algorithms for detection of artefacts, increasing the accuracy of their blood pressure monitors. This could be done with the help of simulators. Considering the pathophysiological characteristics of neonates and infants, the systematic differences between the five monitors have to be regarded as serious.