Because of the different duration criteria to define apnea, and the different methods used to measure obstruction, (either presence of respiratory efforts or absence of cardiac airflow oscillations), it has been difficult to properly assess the prevalence and significance of the various types of apnea. We decided, therefore, to examine the prevalence of central, obstructive and mixed apneas as a function of duration of the respiratory pause in our large database. A total of 12,115 apneas ≥ 3 s collected over 6 years in 86 infants [birthweight 1620±86 g; study weight 2010±111 g; gestational age 31±1 wks; postnatal age 31±3 d] were analyzed. These infants were studied using a flow-through system to measure respiratory pattern and ventilation. Apneas were grouped by duration into 5 epochs yielding the following distribution: 3-5 s [96% central, 3% obstructive; 1% mixed], 5-10 s [91% central, 4% obstructive, 5% mixed], 10-15 s [81% central, 3% obstructive, 16% mixed], 15-20 s [48% central, 11% obstructive, 41% mixed], ≥ 20 s [29% central, 14% obstructive, 57% mixed]. Mixed apneas consisted mostly of an initial central pause followed by an obstructive component (87%). The distribution of apneas using a contingency table analysis showed that the prevalence of obstructive apneas was lower than expected in quiet sleep (85 vs 138; p < 0.001) and higher in indeterminate sleep (95 vs 66; p < 0.02); the prevalence of mixed apneas was higher in REM sleep (204 vs 161; p < 0.02). The findings suggest: 1) the predominance of central or mixed apneas depends entirely on the duration of apnea, short apneas being predominantly central and long apneas (>20s) predominantly mixed; 2) mixed apneas almost always begin as a central apnea; 3) obstructive apneas are rare and even in pauses greater than 20 s they are largely outnumbered by central or mixed apneas; their unusually high prevalence in indeterminate sleep may imply and underlying physiologic mechanism distinct from that of mixed apneas; 4) since most apneas begin as central it is not surprising that, in most instances, they are successfully treated with central respiratory stimulants such as methylxanthines. Supported by The Children's Hosp. Research Found.