Several recent investigations have reported associations between a reduced capacity to fully methylate inorganic arsenic and increased susceptibility to arsenic-caused cancer. In these studies, methylation patterns were based on a single assessment of urinary arsenic metabolites collected at the time of cancer diagnosis. However, the latency of arsenic-caused cancer may be several decades, and the extent to which a recent measurement can be used to estimate a person's past methylation pattern is unknown. In this investigation, the distribution of urinary inorganic arsenic, monomethylarsonate, and dimethylarsinate was used to assess intraindividual variation in methylation capacity in 81 subjects with low to moderate arsenic exposures. Multiple urine samples were collected from each subject over a 1-year period. Duplicate analyses done on 27 samples were used to assess laboratory measurement imprecision. The intraclass correlation coefficients (ICC) for the proportion of urinary arsenic as inorganic arsenic, monomethylarsonate, and dimethylarsinate in samples taken an average of 258 days apart, were 0.45 [95% confidence interval (95% CI), 0.23-0.63] 0.46 (95% CI, 0.24-0.64), and 0.49 (95% CI, 0.28-0.66). In analyses of duplicate samples, ICCs for the concentration of arsenic species ranged from 0.87 to 0.93, whereas ICCs for species proportions ranged from 0.63 to 0.76. These data suggest that individual methylation patterns remain fairly stable over time, although variability due to measurement imprecision or intraindividual changes over time does occur. This variability could lead to misclassification of methylation patterns and could bias relative risk estimates in studies of methylation and cancer towards the null.