The possibility that patients with infantile nystagmus achieve spatial constancy by sampling the visual scene only during certain range of velocities or certain phases of their nystagmus cycle was investigated by asking patients to detect a flashed test target that was presented repeatedly during all phases of the nystagmus cycle. After observing a 543 nm fixation spot projected on a diffusely illuminated tangent screen 1 m in front of the eye, patients were asked to detect a 2 msec test flash of the spot, now locked to the retina, that occurred 200 msec after the fixation spot was extinguished. The test target appeared randomly at 3,6,9, or 12 o'clock at, in separate trials, 0.8 deg or 10.0 deg from central vision. To avoid forcing patients from guessing a direction cued by the disappearance of the fixation spot, sometimes the flash did not occur at all. To avoid dark adaptation in our patients, the diffuse background illumination was adjusted to the brightest level allowing easy detection of the test flash. In every condition, all patients reported the fixation spot stationary. The probability of detection was the same across all velocity ranges. Remarkably, patients were just as likely to detect the test flash when the eye was nearly stationary even when it was moving more than 100 deg/sec. In one patient, the background illumination was raised so that he began missing the test flash more often. Here too, the probability of detection, now reduced to about 50%, was the same across velocity ranges. In some patients we tried to backward mask the test flash by having the fixation spot reappear at various time after the test flash (40-2000 msec). Detection probability was unaffected by the reappearance of the fixation spot with the time interval tested. We conclude that the absence of oscillopsia in our patients was not accomplished by a sampling or masking process.