The dinoflagellate Karlodinium armiger has a huge impact on wild and caged fish during blooms in coastal waters. Recently, a new toxin, karmitoxin, was chemically characterized from K. armiger and a quantification method was established, thereby allowing investigations of the fish killing mechanism. K. armiger is not able to grow in standard growth media that are based on nitrate as a nitrogen source, and successful cultures of this species have only been achieved in mixotrophic cultures after addition of a prey source. Here we show that addition of ammonium (up to 50 µM) to the growth media is a good alternative, as K. armiger batch cultures achieve growth rates, which are comparable to growth rates reached in mixotrophic cultures. Karmitoxin production (1.9 and 2.9 pg cell-1 d-1) and cellular karmitoxin content (8.72 ± 0.25 pg cell-1 and 7.14 ± 0.29 pg cell-1) were in the same range, though significantly different, in prey-fed cultures and monocultures supplied with ammonium, respectively. Net production of karmitoxin stopped when the K. armiger cultures reached stationary growth phase, indicating no accumulation of karmitoxin in cells or growth media. Toxicity tests towards sheepshead minnow fish larvae indicated the rapid death of the fish larvae when exposed to high K. armiger cell concentrations (LT50 of 2.06 h at 44.9 x 103 cells mL-1 cultivated with ammonium). Purified toxins caused the same physical damage to fish larvae as living K. armiger cultures. An exposure of purified karmitoxin to fish larvae and rainbow trout gill cells indicated that the fish larvae were about three times less sensitive than gill cells. When comparing the effect of purified toxins with the effect of whole K. armiger cultures, twice the toxin concentration of the purified toxins was needed to cause the same effect. Although a loss of karmitoxin of twenty percent was observed during the incubation, this could not explain the apparent discrepancy. Other factors, like a direct effect of the K. armiger cells on the fish larvae or other, yet unknown toxins may influence the effect of whole cell cultures. To study the effects of released karmitoxin, fish larvae were exposed to a K. armiger culture that was treated with HP-20 resin, which adsorbs extracellular karmitoxin. The 24 h HP-20 treatment resulted in a K. armiger culture that had 37% less total karmitoxin, without a reduction in cell concentration, and a reduced toxic effect was observed in the HP-20 treated culture, as compared to non-treated controls. Fish larvae that were exposed to HP-20 treated culture were immobilized, but survived during the 12 h exposure, whereas the exposure to non-treated culture led to high mortality of the fish larvae. Direct observations under the microscope revealed no evidence of micropredation of K. armiger on the fish larvae during any of the exposures. Thus, the results presented here, indicate that released karmitoxin is the main cause for fish kills by K. armiger. Finally, we found that juvenile rainbow trout were six times more sensitive than fish larvae towards K. armiger, indicating that juvenile fish are more sensitive to K. armiger in bloom situations than early larval stages.