Abstract One of the remaining challenges in image registration arises for multi-modal images taken from different imaging devices and/or modalities. Starting in 1995, mutual information has shown to be a very successful distance measure for multi-modal image registration. However, as it is well-known, mutual information also has a number of drawbacks. Here, we present an alternative image similarity measure which is also capable to handle multi-modal images but better suited for optimization. As we show, the alternative normalized gradient based approach is deterministic, much simpler, easier to interpret, fast and straightforward to implement, faster to compute, and also much more suitable to optimization.