The ability of human subjects to judge the duration of short empty time intervals was studied in relation to the modality composition of the marker signals. At each trial, a pair of empty intervals was presented by a series of three successive stimuli, and the subject was asked to point out the longer interval of the two. Tone pips and flashes of light were used as the bounding signals. All the possible combinations of auditory and visual stimuli were used, in random order, to delimit pairs of intervals. Performance was found modality-independent when the first two stimuli were of the same modality. Strong response biases were introduced by varying the modality of the first or the second stimulus. Analysis of these biases indicates that memorization of the empty time intervals is affected by the modality of the binding signals.