Recently, random generation of time intervals has been proposed as a proced
ure to impair executive processing in a dual-task paradigm without substant
ial interference with phonological and visuo-spatial working memory resourc
es. A fundamental assumption of this procedure is that humans are able to d
istinguish time sequences on their degree of randomness. The present study
tests this assumption. To this end, nonbiased, repetition-biased, alternati
on-biased sequences and repetitive rhythms were judged under conditions of
higher or lower executive load. In Exp. 1, load depended on the presentatio
n speed, while in Exp. 2, a dual-task condition was used with either a conc
urrent number-copying task or an arithmetic task. It was found that the par
ticipants could distinguish repetitive rhythms from more or less random seq
uences, and that both accuracy of this judgment and latency were affected b
y the concurrent load. The findings are taken as a first indication that ra
ndom time judgment is capacity-limited.