We first reported here that the harmful alga Cochlodinium polykrikoides, which had been previously known as an autotrophic dinoflagellate, was a mixotrophic species. We investigated the kinds of prey species and the effects of the prey concentration on the growth and ingestion rates of C. polykrikoides when feeding on an unidentified cryptophyte species (Equivalent Spherical Diameter, ESD = 5.6 μm). We also calculated grazing coefficients by combining field data on abundances of C. polykrikoides and co-occurring cryptophytes with laboratory data on ingestion rates obtained in the present study. Cocholdinium polykrikoides fed on prey cells by engulfing the prey through the sulcus. Among the phytoplankton prey offered, C. polykrikoides ingested small phytoplankton species that had ESD's ≤ 11 μm (e.g. the prymnesiophyte Isochrysis galbana, an unidentified cryptophyte, the cryptophyte Rhodomonas salina, the raphidophyte Heterosigma akashiwo, and the dinoflagellate Amphidinium carterae). It did not feed on larger phytoplankton species that had ESD's ≥ 12 μm (e.g. the dinoflagellates Heterocapsa triquetra, Prorocentrum minimum, Scrippsiella sp., Alexandrium tamarense, Prorocentrum micans, Gymnodinium catenatum, Akashiwo sanguinea, and Lingulodinium polyedrum). Specific growth rates of C. polykrikoides on a cryptophyte increased with increasing mean prey concentration, with saturation at a mean prey concentration of approximately 270 ng C ml⁻¹ (i.e. 15,900 cells ml⁻¹). The maximum specific growth rate (mixotrophic growth) of C. polykrikoides on a cryptophyte was 0.324 d⁻¹, under a 14:10 h light-dark cycle of 50 μE m⁻² s⁻¹, while its growth rate (phototrophic growth) under the same light conditions without added prey was 0.166 d⁻¹. Maximum ingestion and clearance rates of C. polykrikoides on a cryptophyte were 0.16 ng C grazer⁻¹d⁻¹ (9.4 cells grazer⁻¹d⁻¹) and 0.33 μl grazer⁻¹h⁻¹, respectively. Calculated grazing coefficients by C. polykrikoides on cryptophytes were 0.001–0.745 h⁻¹ (i.e. 0.1–53% of cryptophyte populations were removed by a C. polykrikoides population in 1 h). The results of the present study suggest that C. polykrikoides sometimes has a considerable grazing impact on populations of cryptophytes.