Crosses were performed to identify the sources of variation in zygote production (via cystocarp production) in Gracilaria gracilis, a red haploid-diploid seaweed. First, because male gametes are short-lived (<6 h), the rate of gamete encounters was evaluated in a time-course experiment. Second, the effect of water motion on gamete encounters was assessed by introducing turbulent eddies in the crossing tank and by comparing fertilization rates with and without this added turbulence. Third, variation due to individual performance was explored by performing multiple-donor crosses using 12 males and 12 females from three populations. Paternity of cystocarps produced in these crosses was determined using microsatellite markers. The results show that cystocarp yield increased with exposure time: fertilization occurred in as little as 15 min after the introduction of male branches into the crossing tank and maximum cystocarp production values were observed at 6 h. There were no significant differences in cystocarp production between the two turbulence levels. On the other hand, cystocarp production was highly influenced by male and female parental identities and to a lesser degree by an interaction between the male and female parents. The variation in cystocarp production according to male and female identity was not due to population origin as there was no difference between intra- and inter-population crosses. Thus non-random mating occurs in controlled conditions and arises from differential performance in G. gracilis. There was a strong deviation from equality of male performance, implicating post-adhesion events and/or male gamete production as important in generating non-random mating. Consequently, non-random mating may play a role in the evolution of mating patterns in G. gracilis.