Pressurized water reactor (PWR) assembly reactivity distributions are inferred from ∼600 in-core flux maps taken during 44 cycles of operation of the Catawba and McGuire nuclear power plants. The reactivity distribution for each flux map is determined by systematically searching for fuel subbatch reactivities that minimize differences between measured and computed 235U fission rates. More than eight million core calculations are used to reduce one million measured signals to a set of ∼2500 experimental fuel reactivities for fuel with up to 55 GWd/T burnup. These measured reactivity changes with depletion can be used to validate computer code systems used for burnup credit. To reduce the effort required to quantify computer code system biases and uncertainties, the measured changes in fuel depletion reactivity have been reduced to a set of experimental PWR lattice benchmarks for the change in reactivity as a function of fuel burnup. Results demonstrate that the uncertainty of hot-full-power (HFP) depletion reactivity of the benchmarks is < 250 pcm up to 55 GWd/T burnup. Oak Ridge National Laboratory's TSUNAMI tools are used to extend HFP results to cold conditions, and reactivity decrement uncertainties increase to ∼600 pcm. These experimental benchmarks provide a basis for quantification of combined nuclide inventory and cross-section uncertainties in computed reactivity decrements. It is demonstrated that flux map data reduction is not sensitive to the analytical tools (CASMO/SIMULATE) employed here, and experimental fuel depletion reactivity decrements and uncertainties are anticipated to be independent of fuel management code system use for the data reduction. For CASMO-based analysis, the HFP reactivity burnup decrement biases are shown to be <250 pcm up to 55 GWd/T burnup, and results show that the historical “Kopp memo” 5% reactivity decrement uncertainty assumption is conservative.