In this article, I put forward a basic philosophical claim: empirical scientific knowledge, that is, knowledge generated in experimental and observational practices, presupposes real causation. My discussion exploits two core notions from the philosophical analysis of scientific experimentation and observation: the aim of realizing object-apparatus correlations and the required control of the relevant interactions between environment and experimental or observational system. The conclusion is that, without the notion of real causation, acquiring epistemically sound empirical knowledge is impossible. Several empiricist objections (...) to this conclusion are discussed and refuted. As a consequence, empiricism faces an unsolvable dilemma: either it cannot account for empirical knowledge or it should accept the existence of unobservable but real causal interactions. (shrink)