Conscious handling of uncertainty is critical in the assessment of chemicals, where the demand of knowledge is high compared to the available amount of resources. Consequently weighing between the use of resources and the acceptance of uncertainty has to be a transparent process and this sets up a challenge in relation to the application of mathematical methods. The maximum entropy principle is a useful paradigm for setting up prior information that is as non-informative as possible. It is shown how the use of partial ordering for deriving ranking results can be considered as a maximum entropy strategy in relation to the model structure uncertainty. In this way it seems possible to solve central parts of the problem of resource optimising risk assessment by combining the precautionary principle and screening before more detailed risk assessment is undertaken. The non-parametric approach of partial order technique further appears to be more transparent than most other ranking methodologies for complex systems. Because in the partial order techniques, no more or less known uncertainty will be hidden in the application of specific weighting factors and functional relationships among the criteria.