Background: Muscle invasive bladder cancer (MIBC) frequently recurs despite radical cystectomy, and new biomarkers are needed to guide treatment. DNA methylation plays a key regulatory role in cancer initiation and progression and has been correlated with prognosis in multiple cancer types. We hypothesized that methylation patterns in cfDNA can predict recurrence after curative cystectomy. Methods: Using a clinically annotated USC bladder cancer tissue archive under an IRB approved protocol, we selected 40 patients whose cancer relapsed within 3 years of radical cystectomy, as well as 46 age-matched and stage-matched patients who did not recur after 3 years of follow-up. cfDNA was extracted from the archival plasma, and samples were pooled for whole genome bisulfite sequencing (WGBS) based on pathologic stage and recurrence status. Differentially methylated regions (DMRs) was identified between recurrence and non-recurrence groups in each stage. Results: Pooled analysis across stages yielded 729 DMRs, of which 487 (67%) were hypomethylated in the pooled recurrence group. Of the 729 DMRs from the pooled analysis, 63.5% were in gene bodies, 17.1% were in promoter regions and 47.9% were at CPG islands. Unsupervised clustering of these DMRs accurately distinguished the recurrence groups. Stage-specific analysis yielded DMR sets ranging from 6412 to 9532 DMRs per stage, with a similar trend of global hypomethylation in the recurrence groups. Basespace Correlation Engine analysis revealed that methylation in our recurrence groups (pooled and by stage) was correlated with bladder cancer methylation in publicly available data and also correlated with expression in the TCGA bladder cancer dataset. Conclusions: WGBS was successfully performed in MIBC using cfDNA extracted from clinically annotated archival plasma. Methylation patterns were correlated with publicly available methylation and expression datasets, and cfDNA hypomethylation pre-cystectomy in our cohort was associated with cancer recurrence. A subset of maximally discriminatory DMRs has been selected to define a methylation signature that is currently undergoing validation as a biomarker predictive of recurrence in MIBC.