This paper discusses the current trend in the commercial real estate market to offer more flexible lease contracts. In this paper we assume one form of flexibility: the right to terminate the lease contract prematurely, i.e. a break-option. The lessor is exposed to more risk in cash-flow when he gives this right to a tenant and should actually be financially compensated for this risk. But how much and how? In The Netherlands it is not customary to ask for a financial compensation for flexibility when entering into a lease contract. It is more likely to apply a risk premium on top of the rent; actual a higher rent to compensate for the risk of flexibility. Determining a reasonable risk premium is not trivial because the lessor does not exactly know how long the tenant will lease. This paper describes a method to determine this risk premium and compare the results of this method with risk premiums that appraisers believe are reasonable. The method has been programmed and can be requested from the authors.We apply the real option theory to determine the height of the risk premium. Based on an example we show what the height of the risk premium should be (according to the model) if a standard lease of five years is terminable at the end of year three and year four with a notice of one month. This is just one complex example; it is possible to appraise all other forms of break-options (flexibility) with this method, but in this paper we leave these out of consideration. The results of our model are compared with the estimations of a group of senior appraisers who were asked to determine a risk premium for this break option for a specific building in Amsterdam. They determined an average of 5% risk premium . Our model determined a risk premium of 4%. Although a difference, but fairly small and explainable. The difference is in the estimation of the volatility, the degree in variation of rental prices. To determine the volatility, we used the IPD Index Netherlands, while we asked appraisers to their insight about one specific object. We probably underestimate the volatility because we could not take the unique risks of the object in consideration. We describe how a user of the model may determine the volatility of an unique object, but we do not elaborate further on this.