The nystagmus response elicited by a velocity step of 90°/s was evaluated in a group of 500 consecutive patients By computer analysis and by measuring the duration (t, in s) of the first phase of the response. A 90% log normal range of 37–55 s (mean 45 s) was established for t directly from the electronystagmogram (ENG) in a group of 12 normal subjects. The computer-based response type ‘G–’, i.e. ‘Gesamtamplitude’ (cumulative eye displacement), G < 480°, yielded a detection rate of labyrinthine defects (LDs) of about 65 % with 35 % false positives and 15 % false negatives. A percentage of only about 25 of LDs was detected, with similar percentages of false detections, when using a criterion based on short t values. The lower yield was partly due to the fact that in some 35% of the cases t could not be measured; in the evaluable cases alone, the detection rate was 60% when using the G criterion and 50% when using the t criterion. It is recommended to measure G directly from the ENG, if t is too short or not available, whenever diminished vestibular responses are suspected.