The occurrence of killer toxins amongst yeasts in Brazilian Riesling Italico grape must was investigated by using the sensitive strain EMBRAPA-26B as a reference strain at 18 degrees C and 28 degrees C. From a total of 85 previously isolated yeasts, 21 strains showed ability to kill the sensitive strain on unbuffered grape must/agar (MA-MB) and 0.1 M citrate/phosphate-buffered yeast extract/peptone/dextrose/agar (YEPD-MB) media both supplemented with 30 mg/l methylene blue. The killer activity of only four yeasts depended on the incubation temperature rather than the medium used. At 28 degrees C, the strains 11B and 53B were not able to show killer action. On the other hand, strains 49B and 84B did not kill the sensitive yeast at 18 degrees C. The killer strain EMBRAPA-91B and a commercial wine killer yeast K-l were employed to examine the sensitivity of the isolated yeasts on YEPD-MB and MA-MB at 18 degrees C. The sensitivity and neutral characteristics of yeasts were shown to be dependent on the medium and the killer strain. Interactions, including K- R(-), K- R(+) and K+ R(+) strains, simultaneously, have revealed that some K- R(+) strains appear to pro tect the K- R(-) strain against the killer toxin. Sensitive dead cells, although to a less extent, also exhibited similar protection. Kinetic studies have shown that the maximum specific growth rates were higher for the 20B YEPD-MB-sensitive strain (mu(max) = 0.517 h(-1)) than for both the 91B (mu(max) = 0.428 h(-1)) and K-1 (mu(max) = 0.466 h(-1)) killer strains. The protective capacity of neutral or sensitive cells that contaminate a fermentation, as well as the higher maximum specific growth rate of sensitive yeasts, besides other factors, may pre-rate the dominance of a killer strain. This protective capacity may also reduce the risk of a sensitive inoculum being killed by wild-type killer yeasts ih open non-sterile fermentation.