Speleothem oxygen isotope (δ18O) records provide key insight into the rate and timing of terrestrial paleoclimate changes during the late Quaternary. However, it can be difficult to deconvolve the δ18O signal into individual components, which include processes related to moisture source, moisture transport, temperature, precipitation amount, infiltration, and the cave environment. We developed a framework that uses triple oxygen isotope distributions in speleothems to refine interpretations of δ18O speleothem records. This framework identifies the influence of dominant processes on δ18O values through time by their characteristic (although not necessarily unique) trends in Δ′17O vs. δ′18O space, where Δ′17O = δ′17O – 0.528δ′18O and δ′xO = ln(δxO + 1). Following Guo and Zhou (2019a), we expect that ‘cave kinetic’ processes (e.g., fast degassing at the drip site, prior calcite precipitation) will drive positive trends between δ′18O and Δ′17O. In contrast, we can identify hydrologic processes from near-horizontal trends that reflect Rayleigh-type meteoric water processes and negative trends driven by changes in evaporation processes at the moisture source region or at the cave site, mineralization temperature, and seasonality in precipitation/infiltration amount. We applied this framework to four western USA speleothems from Cave of the Bells (Arizona), Leviathan Cave (Nevada), and Lehman Caves (Nevada). The Cave of the Bells and Leviathan data have near-horizontal to negative trends indicating δ18O variability was driven largely by changes in Rayleigh distillation of atmospheric moisture and moisture source conditions, supporting prior interpretations. We analyzed two Lehman Caves records because they were likely influenced by non-equilibrium processes and the data show weak to moderate negative trends. For sample LMC-12b, chosen for its extreme 7.5‰ δ18O range, the trend is statistically distinct from the near-horizontal Rayleigh-process trend and most consistent with changes in local evaporation intensity and infiltration seasonality as primary drivers. None of these records displays a positive covariation slope between δ′18O and Δ′17O, suggesting limited variability in cave kinetic processes through time or unknown limitations to the kinetic model of Guo and Zhou (2019a). Additionally, reconstructed formation waters for all sites fall near the Δ′17O vs. δ′18O Local Meteoric Water Line, a correlation we suggest as a novel test of the absolute magnitude of isotopic offset due to cave kinetic processes. More broadly, our framework adds context to the only other study of carbonate speleothem triple oxygen isotope composition (Sha et al., 2020). We find that positive to negative Δ′17O vs. δ′18O trends likely exist in speleothem data that may reasonably be expected from regional climate processes and that, combined with other proxy data, triple oxygen isotope data will be useful in constraining interpretations of δ18Ospeleothem records.