Saccadic eye movements provide important information about the nervous system at several levels. Latency, amplitude, duration, velocity and deviation are variables of saccades which are investigated ordinarily. The saccades onsets and ends are localized, typically, using threshold approaches based on the differentiated signal and on the signal itself. This paper present a new algorithm to estimate the onsets and ends of saccades, based on the calculation of the area of a triangle. The performance of this algorithm (in presence of noise) with two based on threshold approaches is performed. The results demonstrate that the proposed algorithm outperforms with the others two in terms of accuracy and robustness to noise on the saccadic electroculographic signal, and it presents the advantage that it doesn't depend on the threshold.