BACKGROUND: The Dutch reimbursement procedure for expensive hospital drugs requires the submission of a baseline cost-effectiveness analysis together with a research plan for the period of temporary reimbursement in order to estimate the real-life cost-effectiveness after 4 years. In this situation, a Value-of-Information (VOI) analysis might identify the critical parameters that need to be studied in such outcome study. OBJECTIVES: To identify when a VOI analysis alongside sensitivity analyses is warranted, and when such VOI analysis will not impact the decision making process. METHODS: We used a hypothetical Markov model with three groups of parameters: costs, utilities and transition probabilities. We studied different configurations of input parameters, forcing the outcomes into different directions on the CE-plane. For each input configuration we performed a multivariate sensitivity analysis (MSA) and a probabilistic sensitivity analysis (PSA). In the MSA, sensitivity was measured as percentage change from baseline INMB. Additionally, we analyzed the expected value of perfect information (EVPI) and the expected value of partial perfect information (EVPPI). Analyses were done for a range of threshold ICERs. RESULTS: For each situation it was possible to predict the shape (but not the absolute value) of the EVPI curve based on the PSA findings. When the PSA plot covered both northern quadrants, MSA and EVPPI came to the same ranking of the groups of parameters. When the outcomes were in the northeast quadrant the ranking differed: MSA indicated costs as most important parameters, for EVPPI this was utilities. When outcomes where in the southwest quadrant, costs were most important in MSA and EVPPI. For both other quadrants, MSA and EVPPI were close to zero for all groups. CONCLUSIONS: Whether MSA and EVPPI come to a different priority setting for future research depends both on the threshold ICER and on the location of results on the CE-plane.