The daily rhythms that adapt organisms to the solar cycle are driven by internal circadian clocks. The hypothesis that the core pacemakers of these clocks consist of auto-regulatory transcriptional/post-translational feedback loops (TTFLs) was first developed in flies and fungi and has now been extended successfully to describe circadian timing mechanisms in mammals and plants. TTFL models revolve around the protein products of 'clock' genes that feedback periodically to regulate their own expression. From this simple beginning, the models have been expanded to encompass multiple, interlinked loops. However, experimental data now highlight the limitations of the TTFL model. Until recently, the focus on transcription caused rhythms in cytosolic signalling pathways to be viewed as outputs of the 'core' transcriptional clockwork, or else as a mechanism for its entrainment by extra-cellular stimuli. Recent work in Arabidopsis thaliana, Drosophila melanogaster and mammals now reveals that cytosolic rhythms in small signalling molecules have a central role within the circadian pacemaker. The logic is consistent across taxa: oscillatory cytoplasmic elements integrate with transcriptional feedback loops to sustain them and determine their rhythmic properties. Thus, clock outputs can constitute inputs to subsequent cycles and so become indistinguishable from a core mechanism. This emphasises the interdependence of nuclear and cytoplasmic processes in circadian pacemaking, such that the pacemakers of some species might encompass the entire cell and its intercellular environment.