Due to demographic developments and the worrying economic situation, the pension system in the Netherlands is under pressure. In order to make it more sustainable, the government proposed several reforms, including the transition to the new pension contract in the second pension pillar. Pension funds and social partners can choose between the nominal contract and the real contract. In the nominal contract, the creation of the buffer implies that pension members are relatively sure of a nominal payment. In the real contract, there is the ambition to index the pension claims in order to keep the purchasing power of members at a constant level. However, the certainty of the buffer is absent. With the transition to one of these two options, more of the financial and longevity risk is shifted to pension members. It seems therefore fair to take into account the risk preferences of the individuals by pension funds in order to make a choice between the nominal and the real contract. The real contract seems to imply a more risky investment mix. The ambition is higher, making that the expected return of investments done should be higher. Therefore, the funds should measure how strong the willingness by their members is to keep their purchasing power at a constant level or whether they prefer the certainty of a nominal payment.Measuring risk preferences is however a difficult task. Normative models are not able to explain observed behavior. As discussed in this paper, individuals show several behavioral biases specific for the pension context, meaning that they are often unable to understand the consequences of their choices. In this paper, we will present some criteria and desirabilities a measurement method should have in order to reflect true risk preferences of members in the pension context. Several risk measurement methods found in literature and practice will be discussed. We will conclude that neither of the methods is satisfying all criteria and desirabilities, but that choice-based approaches show more possibilities than direct attitudinal ways of measuring. The possibility to make it interactive and to show members the consequences of their choices contributes to that. However, the threshold to participate in interactive tools like the distribution builder might be rather high for members. As will be discussed, when building in some degree of paternalism the distribution builder might be more practical to use, without losing its attractive elements.