Perimetry suffers from considerable variability of results. A new technique for estimating individual criterion levels in normal subjects exposes the important role of psychological variables. The analysis depends on features peculiar to acuity perimetry. It can be accommodated in the regular examination. The results can be used to compensate for practice effects in serial examinations and to tighten the range of normal limits. The reduction in variation between normal subjects amounted to 57 per cent.