Tracing produced water origins from wells hydraulically fractured with freshwater-based fluids is sometimes predicated on assumptions that (1) each geological formation contains compositionally unique brine and (2) produced water from recently hydraulically fractured wells resembles fresher meteoric water more so than produced water from older wells. These assumptions are not valid in Williston Basin oil wells sampled in this study. Although distinct average ²²⁸Ra/²²⁶Ra ratios were found in water produced from the Bakken and Three Forks Formations, average δ²H, δ¹⁸O, specific gravity, and conductivity were similar but exhibited significant variability across five oil fields within each formation. Furthermore, initial produced water (“flowback”) was operationally defined based on the presence of glycol ether compounds and water from wells that had produced <56% of the amount of fluids injected and sampled within 160 days of fracturing. Flowback unexpectedly exhibited higher temperature, specific gravity, conductivity, δ²H, and δ¹⁸O, but lower oxidation–reduction potential and δ¹¹B, relative to the wells thought to be producing formation brines (from wells with a produced-to-injected water ratio [PIWR] > 0.84 and sampled more than 316 days after fracturing). As such, establishing an overall geochemical and isotopic signature of produced water compositions based solely on chemical similarity to meteoric water and formation without the consideration of well treatments, well completion depth, or lateral location across the basin could be misleading if these signatures are assumed to be applicable across the entire basin. These findings have implications for using produced water compositions to understand the interbasin fluid flow and trace sources of hydraulic fracturing fluids.