The importance of active mate choice as a component of sexual selection remains controversial, despite recent studies identifying traits that correlate with male mating success. We analyzed the sampling sequences of individual female Guianan cock-of-the-rock, providing quantitative documentation of active female choice in this lekbreeding species. We compared observed sampling patterns with null models of random female movement and of movement weighted by the observed distribution of male success. The courts actually visited by individual females were significantly more clumped and centrally located than predicted by computer simulations of either random or weighted movement. Contrary to simplified hypotheses of female choice, hen cock-of-the-rock were not limited either to sequential comparisons or to threshold decision rules when selecting mates. Instead, they repeatedly sampled a subset of potential mates, using the pool-comparison tactic. Within this pool, females preferentially mated with the males ranked highest in terms of total courtship visits received, and rejected low-ranking males. Our results suggest complex interactions between patterns of female sampling and male dispersion in lek mating systems, and also provide insights into the constraints on female sampling in lek species.