There are several different approaches for error handling in C++ and in this talk I would like to compare them, point out their (dis)advantages and how to get the best from them. I will also show some examples from our avast client repository which misuse these types of approaches. Finally, I would like to talk about a possible way of error handling in the future.