Information Collection Rule (ICR) data were analyzed to investigate relationships between water quality and treatment processes and occurrence of disinfection byproducts (DBPs) in drinking water. A new metric developed to quantify extent of halogen substitution in different byproduct classes indicated strong interclass correlations among bromine fractions in dihaloacetic acids (X2AAs), trihalomethanes (THMs), trihaloacetic acids (X3AAs), and dihaloacetonitriles (X2ANs). These measurements were sensitive to censored data handling. Bromine fraction covariance properties were applied in a test for multivariate outliers to identify data entry or analytical errors which was used in database screening. Database screening indicated a high level of ICR data quality. Recovery of categorical descriptors substantially amplified the data set. Data patterns showed expected relationships between source water quality and disinfection practices. Plants with high organic precursor concentrations preferentially employed chloramines and avoided prechlorination. Plants with high bromide levels also tended to employ chloramines although bromide occurrence did not impact prechlorination practice. Variability in applied chlorine dose among ICR plants diminished when dose was normalized to total organic carbon (TOC); the median chlorine to TOC ratio was 1.54 mg Cl2/mg C. Multiple linear regression models for finished water DBP concentrations at chlorine plants indicated significant shifts across compound classes in the direction and magnitude of influence for bromide, alkalinity, pH, chlorine consumed, and organic precursor concentrations. Results suggested that alkalinity serves as an indicator of organic matter hydrophobicity and reactivity towards DBP formation. pH effects were in accord with current understanding, though observed large differential impacts across species within THM and X3AA classes were not previously noted. Model results suggested that chlorine consumed after initial dose is less relevant for X2AA formation than for other DBPs examined. Based on model projections, use of alternative disinfectants in combination with subsequent chlorination led to substantially lower DBP concentrations compared to use of free chlorine alone. Chloral hydrate, an exception, was enhanced under certain ozone treatment conditions. Model projections indicated that softening treatment led to substantial reductions of brominated THM and X2AA species, and all X3AA species, though total organic halogen was unaffected. Softening effects were attributed to improved organic precursor removal.