The variability of species sensitivity distribution (SSD) due to contaminant bioavailability in soil was explored by using nickel as metal of concern. SSDs of toxicity test results of Avena sativa L. originating from different soils and expressed as total content and available (0.01 M CaCl2) extractable concentration were compared to SSDs for terrestrial plants derived from literature toxicity data. Also the 'free' nickel (Ni2+) concentration was calculated and compared. The results demonstrated that SSDs based on total nickel content highly depend on the experimental conditions set up for toxicity testing (i.e. selected soil and pH value) and thus on metal bioavailability in soil, resulting in an unacceptable uncertainty for ecological risk estimation. The use in SSDs of plant toxicity data expressed as 0.01 M CaCl2 extractable metal strongly reduced the uncertainty in the SSD curve and thus can improve the ERA procedure remarkably by taking bioavailability into account.