Aims: This study explores if nightly hypoxia (i.e. percentage of sleep time with oxygen saturation lower than 90% (SaO₂<90%)) contributed to the association between sleep-disordered breathing (SDB) and insomnia in community-dwelling elderly with and without cardiovascular disease (CVD). A second aim was to explore a potential cut-off score for hypoxia to predict insomnia and the association of the cut-off with clinical characteristics and cardiovascular mortality. Method: A total of 331 community-dwelling elderly aged 71-87 years underwent one-night polygraphic recordings. The presence of insomnia was recorded by a self-report questionnaire. The presence of CVD was objectively established and mortality data were collected after three and six years. Results: In both patients with CVD (n=119) or without CVD (n=212) SDB was associated with hypoxia (p<0.005). Only in the patients with CVD was hypoxia associated with insomnia (p<0.001) which mediated an indirect effect (p<0.05) between SDB and insomnia. Hypoxia of more than 1.5% of sleep time with SaO₂<90% was found to be a critical level for causing insomnia. According to this criterion 32% (n=39) and 26% (n=55) of those with and without CVD had hypoxia, respectively. These groups did not differ with respect to age, gender, body mass index, diabetes, hypertension, respiratory disease or levels of SDB. However, in the CVD group, hypoxia was associated with cardiovascular mortality at the three-year follow-up (p=0.008) and higher levels of insomnia (p=0.002). Conclusion: In the elderly with CVD, SDB mediated by hypoxia can be associated with more insomnia and a worse prognosis.