Estes and Steinberg hypothesized that the intensity of sublittoral marine plant—herbivore interactions among different temperate regions has differed over the past 5—10 million yr, due to differences in the influence of diving mammals that prey on marine herbivores. They argued that because diving predators have not occurred in temperate Australasia, densities of sublittoral herbivores have been historically higher there than in comparable North American systems. This has resulted in marine algae and herbivores in temperate Australasia having strong selective effects on each other. Estes and Steinberg predicted that one result of these strong interactions would be that invertebrate herbivores in temperate Australasia should have evolved greater tolerance to brown algal phlorotannins (polyphenolic compounds that are the dominant secondary metabolites in temperate brown seaweeds) than similar herbivores in North America. We tested this prediction for five invertebrate herbivores in temperate Australasia. Most work was done on three herbivores from New South Wales, Australia, the sea urchins (echinoids) Tripneustes gratilla and Centrostephanus rodgersii, and the gastropod (snail) Turbo undulata. Additional experiments were done using Evechinus chloroticus (urchin) and Cookia sulcata (gastropod) from New Zealand. Both the short—term effects of phlorotannins on feeding behavior, and the longer term effects of phlorotannins on the growth or mortality of herbivores were assessed by (a) feeding the herbivores an array of algal species that varied in their phenolic content, and (b) extracting mixed fractions of phlorotannins from the algae and using these extracts to experimentally manipulate the phlorotannin content of palatable agar disks. The phenolic content of algae used in feeding experiments in Australia was sampled seasonally for 3 yr, and the algae were divided into two groups. These included a phenolic—rich group of five species with a mean phenolic content = 7.02% dry mass (range of species means 6.6—7.2%), and a phenolic—poor group of four species with a mean of 2.69% (range of means = 2.2—3.1%). In strong contrast to earlier results from North America, no Australian herbivore consistently preferred phenolic—poor species over those rich in phenolics. Algal phenolic levels also were not correlated with feeding preferences of the New Zealand herbivores. The Australian herbivores were generally not deterred by the addition of extracted phlorotannins to agar disks at concentrations of 5 mg/mL (13.5% dry mass) in the disks. However, some extracts deterred some herbivores, indicating that the effects of phlorotannins on the herbivores are variable. The effects of phlorotannins at higher concentrations (8 mg/mL or 18.5% dry mass) were equivocal. Deterrence was increased in some instances, but elevated concentrations of phlorotannins from Carpophyllum maschalocarpum enhanced feeding by Tripneustes gratilla. Algal phenolic levels were not correlated with the growth of juvenile Tripneustes gratilla raised on a variety of algal diets for 7 mo in the field. Conversion efficiencies of algal biomass to urchin biomass by these echinoids also did not covary with algal phenolic content. Growth of T. gratilla on agar diets for 7 mo was not affected by the addition of phlorotannins extracted from Sargassum vestitum to the agar at a concentration of 5 mg/mL. Growth of the gastropod Turbo undulata over 9.5 mo on selected algal diets was fastest for one measure growth for snails fed a diet of the phenolic—poor Sargassum linearifolium, but algal phenolic levels were not correlated with growth of T. undulata for two other measures of growth. Moreover, conversion efficiencies of Turbo undulata did not vary with diet, indicating that any differences in growth among different diets were due to differences in consumption rather than some physiological effect. This last conclusion was supported in an experiment where T. undulata was grown on agar diets. Growth of the snails was not affected by the addition of phlorotannins from Sargassum vestitum or Ecklonia radiata to agar diets, and in this experiment the amount of different agar diets consumed did not differ. These results do not support proposals that algal phlorotannins function as generally effective "digestibility reducers" against marine invertebrate herbivores. The pattern of geographical variation in the production of phenolics by brown algae, and the responses of marine herbivores to these compounds, are also not easily explained by current theoretical models such as those based on plant apparency or resource availability. A number of the algae studied here also contained smaller nonpolar secondary metabolites such as terpenes or lactones. In general, these algae were the least palatable to the herbivores, and the herbivores grew most slowly when fed monospecific diets of algae rich in nonpolar secondary compounds. The one nonpolar compound that was tested experimentally, the terpine geranylacetone extracted from Cystophora moniliformis, significantly deterred feeding by Tripneustes gratilla. We suggest that nonpolar compounds will generally have stronger effects against Australasian herbivores than do phlorotannins.