Pulsating illumination of VDU is usually regarded as an important source of visual discomfort or fatigue especially during reading. To understand the causes of these effects, previous studies investigated saccades to stimuli that are presented intermittently on cathode ray tubes at refresh rates of 50 - 125 Hz. Results showed that flicker seems to affect saccadic control both on latencies, saccade extent and velocities. Several explanations have been also mentioned stipulating either a direct effect on saccadic computation during latencies (Kennedy et al., 1998) or during the saccade itself (Baccino, 1999). Moreover, the lack of any DC control condition in these studies does not allow to disentangle between these interpretations. To extend these researches we measured saccades executed from central fixation to eccentric (5 deg) targets (red points of laser light, unmodulated in time). These were presented on a bright background produced by a purpose-made fluorescent lamp that was able to produce either steady DC light, or short pulses of light, since the special fast phosphor had a 2 ms rise-time and a 1 ms fall-time. The moment in time when the saccades began and ended was detected on-line and triggered the lamp to switch between steady light and trains of Gaussian-like light pulses of 4 ms duration. We gave trains of pulses with 2 frequencies (50, 100 Hz) in 4 conditions: during fixation of the central launching position only (1), during the latency only (2), during the latency and during the saccade (3) or during the flight of the saccade only (4). Before and after these trains of pulses, steady light of the same luminance of 67 cd/m2 was presented. Moreover, one condition was completely flicker-free. Compared to a reference condition of continuous DC light, the saccadic latency was significantly longer (F(1,21) = 22.92, p < 0.001, n = 24) when a train of light pulses (at 50 Hz) was presented only during the latency, while steady light appeared before and after. In this case, the saccade onset seemed to be delayed. These findings allow to claim that the main effect of pulses on saccadic control occurs mainly during the computation of the saccade (e.g., during the latency). All other conditions had similar latencies as continuous DC light. condition. Trying to see whether these effects might be due to a kind of phase-locked mechanism operating at 50 Hz, the results failed to show this hypothesis.