Societal concerns about the widespread use of flaring of waste gases have motivated methods for predicting combustion efficiency from industrial flare systems under high crosswind conditions. The objective of this paper is to demonstrate, with a quantified degree of accuracy, a prediction procedure for the combustion efficiency of industrial flares in crosswind by using large eddy simulations (LES). LES is shown to resolve the important mixing between fuel and entrained air governing the extent of reaction to within less than a percent of combustion efficiency. The experimental data from the 4-in. flare tests performed at the CanmetENERGY wind tunnel flare facility were used as experimentally measured metrics to validate the simulation with quantified uncertainty. The approach used prior information about the models and experimental data and the associated likelihood functions to determine informative posterior distributions. The model values were subjected to a consistency constraint, which requires that all experiments and simulations be bounded by their individual experimental uncertainty. The final result was a predictive capability (in the nearby regime) for flare combustion efficiency where no/sparse experimental data are available, but the validation process produces error bars for the predicted combustion efficiency.