Retests in static perimetry.

A large pool of visual field data measured with the JO program on the Octopus automated perimeter was used to compare the outcomes of simulated selective retests to those of global retests. Compared to global retesting, with selective retesting an increase in the mean sensitivity and a decrease in the number of pathological test locations were found on the average. We conclude that global retests should be preferred to selective ones so that the perimetric results are not falsified.