Transcriptional repression is essential for the conversion of crude maternal gradients into sharp territories of tissue differentiation in the Drosophila embryo. Evidence will be presented suggesting that some of the embryonic repressors function through a short-range 'quenching' mechanism, whereby a repressor works over short distances (ca. 50 b.p.) to block neighbouring activators within a target enhancer. This type of repression can explain how different enhancers work autonomously within complex modular promoters. However, at least one of the repressors operating in the early embryo works through a long-range, or silencing, mechanism. The binding of a silencer to a given enhancer leads to the inactivation of all enhancers within a complex promoter. The analysis of chromatin boundary elements suggest that silencers and enhancers might work through distinct mechanisms. We speculate that silencers constrain the evolution of complex promoters.