Analytical data are presented for Cl, Br and I on a regional scale for the Milk River aquifer. The three halides show strikingly similar spatial distributions and are highly correlated. Concentrations are low in the freshwater portions of the aquifer but increase by as much as two orders of magnitude along the margins. However, halide ratios reach nearly constant values moving down-gradient, suggesting the dominance of a common subsurface source for these ions. Ratios of Cl/I and Cl/Br are less than those of seawater and fit an origin derived from the diagenesis of organic matter in the sediments. Halide ratios rule out leakage and/or diffusion from the underlying Colorado Group as a major influence on the chemistry; the favored hypothesis is altered connate seawater diffusing from low-permeability units within the Milk River Formation as the primary source of salts. This hypothesis of an internal source has important implications for solute sources in other aquifers affected by saline waters because it does not require the importation of a distant fluid. The 129I/I ratio has a meteoric value in groundwater collected near the recharge area, but ratios for downflow waters are only 8-70% of this value. Due to the 16 Ma half-life of 129I, these data indicate that most of the increase in dissolved I cannot derive from concentration of a meteoric source by ion filtration, but must have a subsurface origin. Concentrations of 129I produced in situ by spontaneous fission of 238U attain measurable levels only in the oldest waters sampled (ages ??? 105 a), in which it may account for nearly 90% of the total dissolved 129I concentration. Water ages based upon 36Cl/Cl data range up to 2 Ma if uncorrected for any dilution by subsurface sources of dead Cl. If one assumes that the subsurface contributions of Cl contribute at least 90% of total Cl in the distal portion, then the 36Cl-based ages are reduced to ??? 1 Ma, somewhat greater than those estimated by hydrodynamic modeling. ?? 1991.