The increasing incidence of toxic cyanobacterial blooms, together with the difficulties to reliably predict cyanobacterial toxin (e.g. microcystins) concentration, has created the need to assess the predictive ability and variability of the cyanobacterial biomass-microcystin relationship, which is currently used to assess the risk to human and ecosystems health. To achieve this aim, we assessed the relationship between cyanobacterial biomass and microcystin concentration on a spatiotemporal scale by quantifying the concentration of cyanobacterial biomass and microcystin in eight lakes over 9 months. On both a temporal and spatial scale, the variability of microcystin concentration exceeded that of cyanobacterial biomass by up to four times. The relationship between cyanobacterial biomass and microcystin was weak and site specific. The variability of cyanobacterial biomass only explained 25 % of the variability in total microcystin concentration and 7 % of the variability of cellular microcystin concentration. Although a significant correlation does not always imply real cause, the results of multiple linear regression analysis suggest that the variability of cyanobacterial biomass and cellular microcystin concentration is influenced by salinity and total phosphorus, respectively. The weak cyanobacterial biomass-microcystin relationship, coupled with the fact that microcystin was present in concentrations exceeding the WHO drinking water guidelines (1 μg L(-1)) in most of the collected samples, emphasizes the high risk of error connected to the traditional indirect microcystin risk assessment method.