I think the main difference is that, in a declarative statement like: x = (if c then a else b) it is guaranteed that only x is affected. With imperative p

I think the main difference is that, in a declarative statement like: x = (if c then a else b) it is guaranteed that only x is affected. With imperative programming, you have to look inside the conditional to find what is affected. if c then: x = a else: y = a You could still make the above snippet work in a declarative way but, as I say, the targets are explicit x, y = (if c then a, y else x, a) If you have function calls involved then the problem is even worse. In an imperative language, the following code could mutate x, y and any other variables arbitrarily (if the scoping rules allow it): x = 0 if c then: f() Whether this difference is enough to describe "if" as not "control flow" in declarative languages, I don't know, but that is not really an interesting question (just a matter of how you define the term). Edit: another difference is that I think declarative languages usually only use immutable types, so changing one variable is guaranteed not to affect another. Contrast with this Python snippet, where mutating x affects y. x = [1, 2, 3] y = [x, x] x.append(4)