How are we to understand causal relations and analysis in social science? This paper takes R. G. Collingwood's writing about causation as its point of departure for the answering of this question. Two different kinds of causal relations are distinguished from pseudo-causality; of the former, one is directly connected to reason, the other to our ability to manipulate the world. Their interconnection and significance are discussed and the conclusions are drawn that (a) causality belongs to the realm of human praxis and that (b) causal analysis proper is well suited for the social sciences. It is further showed that some important explanations are not causal in any of the above-mentioned senses. These explanations could conceivably be called interpretative descriptions, but it is suggested that perhaps they can be understood as examples of causa sui, of something selfcaused.