Risk assessment provides a key input for determining the need for and extent of remedial actions necessary for sites contaminated with naturally occurring radioactive material or nuclear legacy sites. The choice of a modelling approach for risk assessment, and the corresponding toolsets should fit the assessment context, taking account of the complexity, and be clearly related to the questions to be addressed in the decision-making process. One of the objectives of Working Group 1 of IAEA Modelling and Data for Radiological Impact Assessments II (MODARIA II) Programme is to perform intermodel comparisons for case studies of selected sites, in particular, to help illustrate the applicability of different models and approaches as inputs to decision-making processes. This intercomparison exercise, which included the analysis of potential consequences on the management strategy for contaminated sites, has been performed for two sites: The former uranium mill tailings facility at Zapadnoe, Ukraine, and the phosphate processing facility at Tessenderlo, Belgium. Several models and computer codes have been used for one or both of these cases: AMBER, GoldSim, NORM And LegacY Site Assessment, Preliminary Remediation Goals (PRG)-dose compliance concentration calculator, and RESRAD-OFFSITE. The assessments explore the implications of using differing assessment frameworks and assumptions, as well as alternative modelling tools, on model outputs and as input for corresponding decisions on remediation strategy. This paper reviews both similarities and differences in the results of assessments performed using these different models. It discusses how different approaches can complement one another to help build confidence in the evidence base underpinning decisions. It also discusses the appropriateness of the different modelling approaches in a given assessment context. In one of the case studies in particular (Tessenderlo case study), the remediation strategy is essentially driven by the contamination of the site with heavy metals, such as cadmium. This has significant consequences on the choice of the most adequate approaches and scenarios for assessing the radiological risk and balancing their relative importance with other impacts. The development of a holistic approach to risk assessment is, therefore, highlighted.