The paper describes a reliable and valid method for the automatic detection of sleep spindles in whole night polygraphy. The recording of a multi-channel EEG during sleep polysomnography was performed in 10 healthy volunteers aged 20-35 years. This objective method should improve the time-consuming and subjective visual evaluation by increasing the accuracy and allowing the calculation of quantitative variables (i.e., frequency and amplitude), thereby facilitating scientific work with quantitative data. An important part of the method is the treatment of artifacts (i.e., muscle and spindle-like alpha activity). Compared to hardware solutions, our software method has the advantage of higher flexibility in regard to artifact identification and usual spindle definitions.