We know that there are articles showing that the automatic analysis of type 3 polygraphy recordings underestimates the result of the RDI with regard to polysomnography and the visual analysis.16 However, in those systems the automatic analysis needs both airflow and oximetry signals to detect an event simulating the visual analysis. In fact, the main disadvantage of the automatic reading systems is the recognition of hypopnea, which has a lower threshold than that for manual reading.17 For that reason, if the sensitivity threshold is low to detect an event, a systematic error will occur, which separates the automatic and the visual RDI. In fact, there are bigger differences between the automatic and the visual RDI, which becomes much larger as the RDI becomes higher.16,18 However, there have also been articles19 showing that when the polygraph analyzes only the airflow obtained by pressure measures with regard to polysomnography, the difference in the RDI and the AHI between the two systems is 0.1 events per hour, with high variability; so, it is a precise measure but with high randomized error, whereas visual reading is precise and has a lower randomized error.19 The airflow automatic analysis of our polygraph works like these last articles. In one validation study15 of our polygraph, which was performed in a sleep unit under continuous supervision, in contrast to our study, the difference in the manual and automatic results were similar to ours (Nuñez et al,15 6.28 events per h; our study, 5.68 events per hour). But the concordance limits obtained for the differences in sampling between the automatic flow signal and the visual RDI were too wide to be used as a control.