Categorical classification of groundwater age based on concentrations of tritium (3H) in groundwater can provide useful information for the assessment and understanding of groundwater resources. These data present a three-part groundwater age classification system for the continental United States based on tritium thresholds that vary in space and time: modern (recharged after 1952), if the measured value is larger than an upper threshold; premodern (recharged prior to 1953) if the measured value is smaller than a lower threshold; or mixed if the measured value is between the two thresholds. Inclusion of spatially-varying that vary geographically on the basis of the location of the sample rather than a single threshold accounts for the observed systematic variation in 3H deposition across the U.S. Inclusion of time-varying thresholds rather than a single threshold accounts for the date of sampling given the radioactive decay of 3H. The efficacy of the three-part classification system was evaluated at national and regional scales. The system was evaluated at a national-scale by classifying samples from 1,788 public supply wells distributed across 19 Principal Aquifers (Tritium_PublicSupply_CONUS.csv and Tritium_CentralValley_CA.csv) and comparing those results with expectations based on hydrogeologic principles. The regional-scale data are from five paired networks of shallow and deep wells (287 wells; Tritium_PairedNetworks_CONUS.csv). As expected, modern groundwater is more prevalent in shallower wells than in deeper wells; in fractured-rock and carbonate aquifers as compared to clastic aquifers; in unconfined areas as compared to confined areas; and in humid climates as compared to arid climates. The results from the three-part classification system were also compared to results for groundwater ages from previously published studies. For 14 previous age-dating studies available for comparison, results from this method compare favorably with the previous studies. This work improves on previous work by developing methods: for choosing post-1952 3H thresholds that minimize the misclassification of modern samples as mixed; choosing a pre-1953 threshold to estimate pre-bomb background concentrations; and adds a mixed category to classify samples that are clearly neither entirely modern nor entirely premodern. As with any tritium-based approach, it can fail when the 3H record in precipitation does not accurately reflect the record of 3H in recharge.