<i>Background:</i> The apnea-hypopnea index (AHI) is frequently used to recognize obstructive sleep apnea (OSA) and to evaluate therapy. <i>Objectives:</i> The aim of this study was to determine the AHI variability during a 10-week period, and to discuss its consequences for diagnosis and therapy evaluation. <i>Methods:</i> Fifteen OSA patients (50.8 ± 11.2 years) underwent four polysomnographic (PSG) recordings, with a mean interval between recordings of 3.3 weeks. <i>Results:</i> No differences were found in the average AHI values of the four PSG recordings (p = 0.985). Nevertheless, pooling all data of the 15 participants yielded a smallest detectable difference for AHI of 12.8. Linear regression between the individual means and standard deviations (SDs) of AHI showed that participants with a higher AHI tended to have a higher SD (p < 0.044). <i>Conclusions:</i> These results suggest a considerable intra-individual variability in AHI recordings. Hence, a single-night recording can only recognize OSA when the AHI lies outside a cutoff band surrounding the AHI cutoff point. AHI variability should also be taken into account when evaluating OSA therapy. In this context, it should be noted that it is mainly the approach that we would like to convey to the reader and not the cutoff values per se.