DNA elements such as genes and their regulatory regions must become accessible for protein binding when transcription is activated, which requires reorganization of the nucleosomes that fold the DNA into chromatin fibers. MNase-seq has been instrumental in uncovering the interplay between gene activity and chromatin organization by mapping the average nucleosome occupancy in populations of cells. However, better mechanistic understanding can be obtained from assays that can map nucleosomes along long strands of DNA at single-molecule resolution and without averaging. Here, we show that the combination of DNA methylation, long-read nanopore sequencing, and a novel nucleosome mapping algorithm based on statistical physics results in precise nucleosome footprinting at the single-molecule level over DNA loci exceeding several tens of kbp. Accurate nucleosome mapping was verified in vitro, using chromatin reconstituted on tandem arrays of nucleosome-positioning elements. Genome-wide application on Saccharomyces cerevisiae grown in different transcriptional conditions revealed large heterogeneity of nucleosome distributions upon transcription activation of the model GAL locus. Moreover, neighboring repeats of the ribosomal transcript RDN1 featured long-range correlations in nucleosome occupancy that we attribute to differential transcriptional activity. This enhanced assay allows for both meta-occupancy analysis as well as in-depth single-fiber comparisons of local chromatin aberrations in the context of transcription, DNA repair, and other processes, illustrating the added value of single-molecule nucleosome mapping using long-read sequencing compared to traditional population-averaged maps.