The marine alga Chattonella marina (Raphidophyceae) has long attracted global attention for its association with massive mortality in wild and cultured fish worldwide. Respiratory disorder is generally believed to be the major cause of fish death from C. marina. Recently, however, osmotic distress as well as suffocation was proposed as a probable cause of fish death from C. marina, but the mechanisms of C. marina induced respiratory and osmotic disorders in exposed fish remain undefined. Moreover, the toxin(s) involved in fish kills from C. marina remains highly controversial. Brevetoxins, free fatty acid, hemolysins, hemagglutinins, reactive oxygen species, and nitric oxide are thought to be the C. marina toxins. However, no concrete data has been provided to substantiate the claims that these are involved in fish kills. The present study aimed to (i) delineate the involvement of respiratory and osmotic disorders in the fish kill mechanisms, and (ii) explore toxins from the harmful alga C. marina. In Part I of this study, it is hypothesized that respiratory disorder is the major cause of fish mortality from exposure to C. marina, and fish susceptibility to C. marina is proportional to their susceptibility to hypoxia which is determined by the adaptability of the fish under hypoxia. Three marine teleosts, the goldlined seabream (Rhabdosargus sarba, sensitive to C. marina), the marine medaka (Oryzias melastigma, semi-susceptible to C. marina), and the green grouper (Epinephelus coioides, tolerant to C. marina) were used to unravel the fish kill mechanisms of this harmful alga. The small model fish marine medaka was employed to determine whether fish tolerance to C. marina is related to their tolerance to hypoxia and/or hyperosmotic stress. Marine medaka of different ages (from 4 month to 8 month old) were exposed to C. marina, hypoxia (mimic respiratory disorder) and 70‰ salinity (mimic hyperosmotic stress). The tolerance of marine medaka to hyperosmotic stress increased with age. Conversely, fish tolerance to C. marina decreased with age. The susceptibility of marine medaka in hypoxia increased with age, as it did in C. marina. The results agree with our earlier findings that the fish species susceptible to hypoxia was also susceptible to C. marina. It is concluded that fish tolerance to C. marina is highly related to their adaptability under hypoxia. The goldlined seabream, susceptible to C. marina, was employed to investigate temporal changes of biochemical, histopathological and physiological parameters related to respiratory function in fish, with an attempt to identify the cause of respiratory disorder induced by C. marina. A significant increase (ca. 40%) of respiratory rate (as measured by frequency of operculum movement) was evident in C. marina exposed goldlined seabream. No significant changes in methemoglobin level and blood lyses rate were detected in C. marina exposed fish, suggesting hemoglobin oxidation and blood lyses were not the key reasons for blood pO2 drop in fish. Gill histopathology such as irregular organization of lamellae and mucous together with algal cells trapped in interfilamental spaces, were typical in C. marina exposed fish. Such changes may impair gas exchange, reduce oxygen uptake and induce hyperventilation in fish. A surge of plasma lactate (indicate anaerobic respiration) occurred in fish shortly after exposure to C. marina (0.5 h) and that persisted throughout the exposure period. Depletion of liver glycogen and plasma glucose were clearly evident in fish showing stress symptoms and near moribund (p < 0.05). A continue metabolism of reserved liver glycogen to supply glucose for anaerobic respiration may explain why tissue ATP levels (in muscle ca 30 times > in liver) remained unchanged in fish exposed to C. marina. Living fish in the C. marina and control fish exposed to non-toxic alga and seawater did not exhibit any significant decline of blood pO2, plasma glucose or liver glycogen levels, even though increases of plasma lactate and osmolality were found in fish alive after C. marina exposure. The green grouper is highly tolerant to C. marina (no mortality after 48 h exposure). The operculum movement rate increased (ca. 50%) in green grouper exposed to C. marina and then declined to basal level after 5 h exposure. Liver glycogen in green grouper exposed to C. marina showed a significant decrease in the first 12 h exposure and returned to control level in the following 12 h exposure. These results suggested green grouper suffer O2 shortage but adapt in C. marina exposure. Metabolic changes including plasma lactate surge, glycogen and glucose depletion that occurred in moribund goldlined seabream were not found in green grouper upon exposed to C. marina. Compared to the goldlined seabream, the green grouper exhibited a 3-times higher level of liver glycogen reserve and a 6-times lower level of basal respiratory rate. A low respiratory rate (operculum movement) means a less frequent contact of fish gills with C. marina cells (and possibly the toxins). This may explain the high survival of green grouper under C. marina exposure. In Part II of this study, the toxic conditions and toxin characteristics of C. marina were studied. The mid-log phase was found to be the most toxic phase based on mortality of marine medaka and goldlined seabream exposed to C. marina at different growth phases. No dose response relationship was found between density of C. marina cells and toxicity to fish. To determine whether the C. marina toxins are water soluble (hydrophilic) or whether direct contact is required to trigger fish kills, goldlined seabream and marine medaka were exposed to C. marina directly and indirectly using a plankton net insert to separate the algal cells from contacting the fish. Mortalities occurred only in treatments where fish were in direct physical contact with C. marina cells. No fish kills were found even at high cell density of C. marina at the most toxic phase where no physical contact was allowed between algal cells and fish. Hydrophilic bioactive compounds that can penetrate through a plankton net are, therefore, unlikely to be the major toxins of C. marina. Cell surface or intracellular lipophilic compounds of C. marina may be involved. Brevetoxins and free fatty acids, suspected as C. marina toxins, are all lipophilic compounds. However, we rejected brevetoxins and free fatty acids as the ichthyotoxins of C. marina by comparing the toxicity of C. marina with a brevetoxins-producing alga Karenia brevis and their organic solvent extracts using the marine medaka. In contrast to C. marina, K. brevis exhibited a very good dose response relationship between cell density and fish mortality. Marine medaka developed significant hyperventilation response to C. marina, but hypoventilation response to K. brevis. Moreover, the organic extracts from C. marina showed no toxicity to fish whereas organic extracts from K. brevis showed significantly higher toxicity than the K. brevis cell culture. The toxins produced by C. marina may be protein in nature or small and labile molecular compounds which are not able to be extracted by traditional organic extraction methods. Alternative approaches (for protein or small molecule compounds) should be adopted or developed in future. On the basis of both the present findings and our previous results on C. marina disruption of epithelial paracellular tight junction, we postulate the fish kill mechanisms of C. marina to be as follows: a direct contact between C. marina cell and gill epithelia is critical to trigger a loosening of paracellular tight junctions in gill epithelia. This may lead to impairment of lamellae integrity and together with mucus secretion, these will further impair gas exchange, reduce oxygen uptake and induce hypoxia in the exposed fish. Fish with a higher tolerance to hypoxia could be more tolerant to C. marina. Fish species having a low basal respiratory rate and high liver glycogen reserve may have a better chance of surviving in C. marina blooms. The middle exponential phase of C. marina showed the highest toxicity to fish and this declined toward the stationary phase. These findings substantiate our earlier report that sub-bloom levels of C. marina can kill fish. Therefore, from the aquacultural and fisheries management viewpoint, setting up regular monitoring of C. marina in fish culture and raising high risk warning signals upon detection of sub-bloom levels of C. marina is advisable. Also, since, the physical contact of fish with C. marina cells is necessary for fish toxicity, it is suggested that, to avoid direct contact of algal cells with fish during C. marina blooms, fish cages can be lowered below the C. marina blooms which are normally concentrated at the water surface during the day. Other strategies to increase fish survival or reduce/avoid fish loss in C. marina blooms include i) increasing the oxygen supply to affected fish, especially at night, and ii) choosing culture hypoxia tolerant fish species in potential C. marina bloom areas.