Generally, the study of gait requires the detection of successive heel contacts and toe-off instants. Traditional gait analysis methods obtain these gait events using dynamometric platforms together with stereophotogrammetric data. Usually, are kept valid only those walking trials where the subjects step on each platform by only one foot. For subjects suffering from walking impairments it is very difficult or sometimes impossible to walk naturally and step properly on the dynamometric platforms. The aim of the present study is to propose a new method to identify, in an automatic manner, the initial contact (IC) and the toe-off (TO) time instants using only stereophotogrammetric data and a classic gait analysis protocol. The assessment of spatio-temporal gait variables during natural walking is also performed. The study consisted in analyzing healthy and Parkinsonian elderly subjects. The reliability of the proposed stereophotogrammetric-based method was tested by direct comparison with the IC and TO instants determined by the dynamometric platform data. The absence of any statistically significant differences between the values estimated by the two different modalities, highlights the reliability of the proposed method in the assessment of these two gait events. Results underline, as expected, the reduction of walking velocity in pathological patients during free ambulation. The present study proposes this method as a valid alternative to the traditional technique that use dynamometric platforms to identify main gait events, for subjects unable to walk naturally and to step properly on the platforms.