The visual census technique is a non destructive method for the estimate of fish populations. The quality of population estimates depends on visibility, structure of the habitat, behaviour of fishes and experience of the diver. Adequate fishing methods for a habitat are less problematic, but entail more logistical organisation for fishing gear, boat and personnel. In this study sampling of seagrass meadow ichthyofauna was conducted with a beach seine and the visual census technique. The different methods of sampling lead to divergent pictures of the seagrass fish community. Beach seining recorded 109 species, whereas visual census only 48 species of fishes. There are also differences in the abundance of the main families. Most abundant families in beach seine catches are Labridae (50,2 %), Apogonidae (15,7 %), Lethrinidae (7,9 %), Scaridae (7,3 %), and Bothidae (5,3 %). Records by the visual census technique showed the following ranking in abundance: Pomacentridae (33,0 %), Siganidae (17,8 %), Plotosidae (16,3 %), Labridae   (15,2 %), and Apogonidae (12,8 %). Fast swimming species such as Siganus canaliculatus (Siganidae) and Cheilio inermis (Labridae) are able to escape out of the slowly pulled beach seine and are underestimated with this method. The visual census technique underestimated the well camouflaged species of the families Labridae and Scaridae and is not able to detect cryptic families such as Bothidae. In this study beach seining provided the best data about the structure of the seagrass meadow ichthyofauna and the visual census technique supported additional data. Both techniques are recommended to investigate the structure of the seagrass meadow ichthyofauna.