During volcanic eruptions, empirical relationships are used to estimate mass eruption rate from plume height. Although simple, such relationships can be inaccurate and can underestimate rates in windy conditions. One-dimensional plume models can incorporate atmospheric conditions and give potentially more accurate estimates. Here I present a 1-D model for plumes in crosswind and simulate 25 historical eruptions where plume height H_obs was well observed and mass eruption rate M_obs could be calculated from mapped deposit mass and observed duration. The simulations considered wind, temperature, and phase changes of water. Atmospheric conditions were obtained from the National Center for Atmospheric Research Reanalysis 2.5° model. Simulations calculate the minimum, maximum, and average values (M_min, M_max, and M_avg) that fit the plume height. Eruption rates were also estimated from the empirical formula M_empir = 140H_obs^4.14 (M_empir is in kilogram per second, H_obs is in kilometer). For these eruptions, the standard error of the residual in log space is about 0.53 for M_avg and 0.50 for M_empir. Thus, for this data set, the model is slightly less accurate at predicting M_obs than the empirical curve. The inability of this model to improve eruption rate estimates may lie in the limited accuracy of even well-observed plume heights, inaccurate model formulation, or the fact that most eruptions examined were not highly influenced by wind. For the low, wind-blown plume of 14–18 April 2010 at Eyjafjallajökull, where an accurate plume height time series is available, modeled rates do agree better with M_obs than M_empir.