In prokaryotes, transcription and translation are coupled. In eukaryotes, by contrast, the spatial separation of the two processes prevents this: mRNA is transcribed in the nucleus, exported to the cytoplasm and only then translated. But are things that simple? Recent work has hinted that some translation might actually take place in the nucleus. Moreover, ribosome components have cropped up in nuclear preparations, and there are indications that surveillance mechanisms that detect faulty mRNAs in the nucleus rely on transcription-translation coupling. Set against this evidence are numerous studies that argue against nuclear translation. So what are we to believe? In a hypothesis on p. 5713, Peter Cook and coworkers put the case for the defence.