Electroencephalogram spindle patterns corresponding to two different phenomena-natural sleep and propofol anesthesia-are compared. The spindles are extracted from 5 overnight sleep recordings and 10 recordings of deep propofol anesthesia. Mean frequency, angle of the trend in instant frequency as well as 3 nonlinear parameters-spectral entropy, approximate entropy, and Higuchi fractal dimension- are calculated to characterize the spindle waveforms. Using the Wilcoxon rank sum test with significance level of 0.01, all the mentioned features, except approximate entropy, differ significantly for the two types of EEG spindles.