Hormones typically induce rapid activation of nuclear transcription factors within minutes after stimulation, and thereby elicit changes in gene expression on a time scale of hours. In my talk, I will discuss such slow gene expression responses and their impact on cell fate decisions. The first part of the talk will be focussed on feedback regulation of the upstream signalling network by the gene expression machinery. Based on stimulus-induced gene expression profiles, it will be shown that all major mammalian signalling pathways are subject to transcriptional negative feedback regulation, and that feedback generally occurs by induction of unstable signal inhibitors. The physiological relevance of transcriptional feedback regulation is then confirmed for specific biological systems (TGFbeta and MAPK signalling) using dynamical modelling and experimental measurements at the protein level. Biological information is frequently encoded in the quantitative aspects of incoming upstream signals (e.g., amplitude and duration). However, both the topology of the downstream transcription factor networks and the actual mechanisms of input decoding at the gene expression level remain unclear. In the second part of the talk, these questions will be addressed by reverse engineering an oncogenic transcription factor network based on siRNA-mediated perturbations and by analyzing the cyanobacterial iron stress response in a combined experimental and theoretical approach.