In this paper we present a correlational study in which we assess the ability of five size measures to predict the number of changes to requirements for a medium size software project. The study is explorative, i.e. we analyse the data collected for our measures to find out the best predictor of number of changes. To our knowledge, no empirical validation of requirements change measures as predictors has been performed in an industrial setting. Based on the data collected from two industrial projects for five measures of size of requirements (number of actors, use cases, words, lines, and revisions), we have built and evaluated prediction models for number of changes to requirements. These models can help project managers to estimate the volatility of requirements and minimize the risks caused by volatile requirements, like schedule and costs overruns.We performed a cross systems validation. For our best model we calculated a pred(0.25)=0.5, which is better than the accuracy of common effort prediction models like for example COCOMO. Although our models are likely to have only local validity, the general method for constructing the prediction models could be applied in any software development company. In an earlier study, we showed that decisions solely based on developer perception are unreliable. Predictions models, like the one presented here can help to mitigate that risk.