Thermally sensitive devices—thermistors—have usually been used to monitor sleep-breathing disorders. However, because of their long time constant, these devices are not able to provide a good characterization of fast events, like hypopneas. Nasal pressure recording technique (NPR) has recently been suggested to quantify airflow during sleep. It is claimed that the short time constants of the devices used to implement this technique would allow an accurate analysis of fast abnormal respiratory events. However, these devices present errors associated with nonlinearities and acoustic resonance that could reduce the diagnostic value of the NPR. Moreover, in spite of the high scientific and clinical potential, there is no detailed description of a complete instrumentation system to implement this promising technique in sleep studies. In this context, the purpose of this work was twofold: (1) describe the development of a flexible NPR device and (2) evaluate the performance of this device when compared to pneumotachographs (PNTs) and thermistors. After the design details are described, the system static accuracy is evaluated by a comparative analysis with a PNT. This analysis revealed a significant reduction (p<0.001) of the static error when system nonlinearities were reduced. The dynamic performance of the NPR system was investigated by frequency response analysis and time constant evaluations and the results showed that the developed device response was as good as PNT and around 100 times faster (τ=5,3 ms) than thermistors (τ=512 ms). Experimental results obtained in simulated clinical conditions and in a patient are presented as examples, and confirmed the good features achieved in engineering tests. These results are in close agreement with physiological fundamentals, supplying substantial evidence that the improved dynamic and static characteristics of this device can contribute to a more accurate implementation of medical research projects and to improve the diagnoses of sleep-breathing disorders.