Assimilation methods for numerical weather prediction models have been highly developed and often rely on interpolation methods that require a minimization of a cost function. Typically these minimizations are accomplished using a least square technique. Often the solution becomes poorly conditioned and iterative solutions are required. This work demonstrates the use of a genetic algorithm for assimilating data. This method is quite robust and does not require least square assumptions. Sensitivity analyses are performed to examine the behavior with other measurement criteria and with changes in genetic algorithm parameters. Examples include assimilating contaminant concentrations and meteorological data into dispersion models.