Abstract: An algorithm of optimal sensor placement for damage diagnosis is discussed in this paper. Taking the maximum off-diagonal element of Modal Assurance Criterion (MAC) matrix as objective function, the minimum of objective function is transformed to the maximum of fitness function. Genetic algorithms are used to optimize the sensor placement. A dualistic coding genetic algorithm is developed to solve a constraint problem in the combinatorial optimization. A numerical example is given to show that the method is feasible in optimal sensor placement of offshore platforms aimed at damage diagnosis. The method is more effective than serial methods.