The relationship between chromatin structure and gene expression is a subject of intense study. The universal transcriptional activator Gal4 removes promoter nucleosomes as it triggers transcription, but how it does so has remained obscure. The reverse process, repression of transcription, has often been correlated with the presence of nucleosomes. But it is not known whether nucleosomes are required for that effect. A new quantitative assay describes, for any given location, the fraction of DNA molecules in the population that bears a nucleosome at any given instant. This allows us to follow the time courses of nucleosome removal and reformation, in wild-type and mutant cells, upon activation (by galactose) and repression (by glucose) of the GAL genes of yeast. We show that upon being freed of its inhibitor Gal80 by the action of galactose, Gal4 quickly recruits SWI/SNF to the genes, and that nucleosome "remodeler" rapidly removes promoter nucleosomes. In the absence of SWI/SNF, Gal4's action also results in nucleosome removal and the activation of transcription, but both processes are significantly delayed. Addition of glucose to cells growing in galactose represses transcription. But if galactose remains present, Gal4 continues to work, recruiting SWI/SNF and maintaining the promoter nucleosome-free despite it being repressed. This requirement for galactose is obviated in a mutant in which Gal4 works constitutively. These results show how an activator's recruiting function can control chromatin structure both during gene activation and repression. Thus, both under activating and repressing conditions, the activator can recruit an enzymatic machine that removes promoter nucleosomes. Our results show that whereas promoter nucleosome removal invariably accompanies activation, reformation of nucleosomes is not required for repression. The finding that there are two routes to nucleosome removal and activation of transcription-one that requires the action of SWI/SNF recruited by the activator, and a slower one that does not-clarifies our understanding of the early events of gene activation, and in particular corrects earlier reports that SWI/SNF plays no role in GAL gene induction. Our finding that chromatin structure is irrelevant for repression as studied here-that is, repression sets in as efficiently whether or not promoter nucleosomes are allowed to reform-contradicts the widely held, but little tested, idea that nucleosomes are required for repression. These findings were made possible by our nucleosome occupancy assay. The assay, we believe, will prove useful in studying other outstanding issues in the field.