This study tested the hypothesis that the snail Hydrobia ulvae will recognise differences in the contamination levels of sediment and avoid the more contaminated zones. Instead of testing avoidance behaviour in a two-compartment system, a linear contamination gradient has been devised in which several zones containing sediment with different contamination levels could be "chosen". A sediment of known severe contamination was collected and mixed with a non-toxic sediment to comprise a linear gradient with four concentrations of 0, 30, 60 and 100% contamination. Thus, the individuals were free to move between fields of different concentration (unforced conditions). During 24 h of exposure, the distribution of the organisms on fields was recorded and organisms' percentage that avoided each field was calculated for each concentration. At 30 and 60% of contamination, avoidance was around 50% for all periods of exposure. Avoidance at 100% contamination ranged between 11 and 56% (rates lower than expected). Organisms that did not avoid the 100% contamination revealed that they were inactive, indicating either that non-mobility and retraction within their shell may be alternative defence strategies, or else the high contamination may have impaired their ability to escape. The percentage of preference shown to the uncontaminated sediment ranged between 64 and 74%. The initial hypothesis was supported: H. ulvae avoids the more contaminated sediment. The assay proposed has been shown to be simple, rapid and effective; in addition, it is considered ecologically useful since the effects resulting from the avoidance are similar to the extinction of the population.