<p>Altered methylation of genes implicated in depression has been proposed as a biomarker for treatment response. Yet, replication remains inconsistent while methodological concerns have grown. In a matched case control study, we investigated DNA-methylation of the <i>NR3C1</i> (glucocorticoid receptor), <i>SLC6A4</i> (serotonin transporter), and <i>OXT</i> (oxytocin)&#xa0;genes in peripheral blood of patients with depression (<i>N</i> = 66) and healthy controls (<i>N</i> = 66) at baseline and after two weeks of inpatient treatment. We tested group differences, convergence over time, associations with changes in depression severity, and relations to mRNA abundance. At baseline, <i>OXT</i> methylation was significantly lower in patients, but this effect disappeared when adjusting for neutrophil and lymphocyte abundance, highlighting the impact of cellular composition on methylation measures. Despite significant decrease in depression severity, methylation changes in none of the candidate genes were associated with treatment response. In healthy controls, two <i>OXT</i> CpG sites showed negative associations with mRNA abundance, which were absent in patients. These findings emphasize the importance of rigorous control of confounders, especially blood cell composition, in psychiatric epigenetics. Our results add to accumulating null findings for stress- and serotonergic-related genes and caution against overinterpreting peripheral methylation signals as biomarkers for depression.</p>