The latencies of on- and off-responses on tonal stimuli have been measured by a pair of electronic clocks. Short responding times have not been forced and interaction between audiometrist and subjects has been perserved. Experienced probands and patients reacted by lengthening of reaction time near threshold and near thresholds in masking noise. In most cases stimuli at threshold levels have not been responded completely. Recruitment is not the only possible cause for short latencies at threshold levels.