The application of an automatic sleep spindle detection procedure allo
wed the documentation of the topographic distribution of spindle chara
cteristics, such as number, amplitude, frequency and duration, as a fu
nction of sleep depth and of recording time. Multichannel all-night EE
G recordings were performed in 10 normal healthy subjects aged 20-35 y
ears. Although the interindividual variability in the number of ; slee
p spindles was very high (2.7 +/- 2.1 spindles per minute stage 2 slee
p), all but two subjects showed maximal spindle activity in centro-par
ietal midline leads. Moreover, this topography was seen in all sleep s
tages and changed only slightly - to a more central distribution - tow
ards the end of the night. On the other hand, slow (11.5-14 Hz) and fa
st (14-16 Hz) spindles showed a completely different topography, with
slow spindles distributed anteriorly and fast spindles centro-parietal
ly. The number of sleep spindles per min was significant depending on
sleep stages, with the expected highest occurrence in stage 2, and on
recording time, with a decrease in spindle density from the beginning
towards the end of the night. However, spindle amplitude, frequency an
d individual duration was not influenced by sleep depth or time of the
night.