Understanding cell-to-cell variability in cytosine methylation is essential for understanding cellular perturbation and its molecular machinery. However, conventional methylation studies have focused on the differences in the average levels between cell types while overlooking methylation heterogeneity within cell types. Little information has been uncovered using recent single-cell methods because of either technical limitations or the great labor required to process many single cells. Here, we report the highly efficient detection of cell-to-cell DNA methylation variability in liver tissue, based on comparing the methylation status of adjacent CpG sites on long sequencing reads. This method provides abundant methylation linkage information and enables genome-wide estimation of cell-to-cell variability. We observed repressed methylation variability in hypomethylated regions compared with the variability in hypomethylated regions across the genome, which we confirmed using public human sperm data. A gradual change in methylation status at the boundaries of hypomethylated regions was observed for the first time. This approach allows the concise, comprehensive assessment of cell-to-cell DNA methylation variability.