Normalizing new data in Artificial Neural Networks

I understand the reasoning behind normalizing data for training ANNs. But once the ANN is trained and we want to run test data against it, how do we normalize the data? For example, consider a SOM. All our training data is normalized to fall between the values 0 and 1. Now take our test data. This data, if normalized, may not fall between 0 and 1. It is possible to get values that are less than 0 or greater than 1. Is there an approved method to do this or do we just take those values as is? Thanks