Populations of herbivorous macroinvertebrates living on dead coral rock substrata were surveyed on three midshelf reefs (Davies Reef, Little Broadhurst Reef and Centipede Reef) and two outershelf reefs (Chicken Reef and Anzac Reefs No. 2). Initial surveys were conducted on Davies Reef in April and October 1988 and were extended to the other four reefs in March 1989.At Davies Reef, six zones located on the reef crest and 25, 160, 40, 270 and 380 m from the reef crest were defined along a transect extending from the windward reef crest to the lagoon. Quadrats (2 m x 2 m) were placed haphazardly over areas of hard substrate within a 20 m x 50 m area parallel to the reef crest in each zone. In April 1988, the substrate within each of a minimum of 8 quadrats per zone was searched and herbivorous macroinvertebrates identified and counted. In October 1988, the densities of the most common grazing gastropods (Turbo chrysostomus, Turbo argyrostomus and Trochus spp.) were measured. Between 10 and 40 quadrats (2 m x 2 m) within each zone were searched during the day and the gastropods indentified, counted and the shell length measured.In March 1989, surveys were extended to four other reefs in the area to compare densities of groups of grazing macroinvertebrates. Between 3 and 6 sites were selected along the windward side of each reef, in the zone approximately 160 m from the reef crest (mid-reef flat). A minimum of 18 quadrats per site were surveyed.Three different techniques were used to estimate the rate of ingestion of epilithic algae by Turbo chrysostomus on Davies Reef in October 1988. Ingestion rates of 34 Turbo chrysostomus were measured using the faeces quantification technique. Individuals were placed in separate containers within 24 hours of collection with a 10 x 10 cm slab of turf covered coral rock. After a further 24 hours, faecal pellets were collected and dry and ash-free dry weight was determined. Subsamples of algae scraped from the slabs of coral rock were analysed for ash-free dry weight and organic carbon. Comparative data for 6 Trochus pyramis was obtained in March 1989. Experiments using the grazer manipulation technique were conducted twice in the zone 160 m from the reef crest. Nine slabs of natural coral rock (each approximately 12 x 12 cm), collected from Turbo chrysostomus habitat were placed individually into wire mesh cages attached to the substrate. Two Turbo chrysostomus were placed in each of 6 caged treatments, with the remaining cages used as controls. After 48 hours, the snails had produced distinct feeding scars on the turf layer and the slabs were collected and frozen. The area of turf covering the upper surface was measured and turf algae on the slabs was removed dried, weighed and subsampled for determination of ash-free dry weight and organic carbon.A total of 210 Turbo chrysostomus were collected from the same zone to determine ingestion rates using the crop-fullness technique. Snails were collected at 2 hourly intervals during the day and night and at 30 minute intervals between 1800 and 2200 hours, the expected crop filling period. The animals were separated from their shells and opercula, after measurement of the shell length, and preserved in 4% formalin. The animals were later dissected and the crop contents collected for determination of dry and ash free dry weight. Subsamples of the crop contents were analysed for organic carbon. Dry weight of the soft tissue of the animals was also determined. Comparative data was collected for 24 Trochus pyramis, collected over day and night in February 1989.Feeding behaviour of Turbo and Trochus was observed to determine the period of feeding activity and the nature of the material ingested. On Davies Reef, observations were made during the day and night and the proportion of specimens that were cryptic or active and feeding recorded. Observations were also made on specimens maintained in an aquarium containing coral rock collected from the natural