We propose an evolutionary approach for target allocation in tactical level land combat. The purpose is to assign friendly military units to enemy units such that the total weapon effectiveness used is minimised while the attrition goals set for the enemy units are satisfied. A repair algorithm is developed to ensure feasibility with respect to the attrition goal constraints. A tightness measure is devised to determine the population size of the genetic algorithm as a function of constraint tightness. Also, a local improvement algorithm is used to further improve the solution quality. Experimental results indicate that the genetic algorithm can find solutions with acceptable quality in reasonable computation time. Although the approach is developed for the target allocation problem, it can be adapted for other assignment problems.