Male and female mating preferences are commonly inferred from association times spent with potential mates in a dichotomous-choice test. However, this assessment method is rarely validated, particularly so for male mating preferences. Using the Trinidadian guppy (Poecilia reticulata), an important model species in the study of sexual selection, we tested whether a male’s mating preference for either of two stimulus females in a dichotomous-choice test predicted his mating behaviours directed at the preferred female when he was allowed to swim freely with both females. First, we presented individual males with two females that differed in body length in a dichotomous-choice apparatus in which the male could only use visual cues to assess the paired females. We quantified male mating preference as the duration of time a focal male spent associating with each female. Immediately following this test, the focal male was allowed to swim freely with both females, and we quantified the time he spent sexually pursuing each female and the number of courtship sigmoid displays and copulation attempts he directed at each female. On average, males did not significantly prefer either of the two stimulus females in either of the two tests; however, the magnitude of male preference for the larger female tended to increase as the size difference between the paired females increased. More importantly, there was a significant positive relationship between male association time in the dichotomous-choice test and both the time spent sexually pursuing and the number of courtship sigmoid displays directed at the same female initially preferred in the dichotomous-choice test. Collectively, these results confirm that association time measured in a dichotomous-choice test is a reliable predictor of male mating preferences in the Trinidadian guppy.