Is Immanuel Kant's distinction between noumenal and phenomenal realms valid?

In the past, I've dismissed Kant's distinction between noumenal and phenomenal realms simply because Kant seems to suggest that the noumenal realm is unknowable. If that were true, then not only can he not say anything about the noumenal realm, but he can't even say that it's unknowable because even that would require some evidence. Is that right? Does the distinction have any validity? If not, why is it so widely accepted by philosophers?