In this paper design of very thin and broadband electromagnetic absorbers is described. For normal incidence a five-layer absorber with 7.6 mm thickness with reflection coefficient below 20 dB over the frequency range 0.1 < f (GHz) < 10 has been achieved. Genetic Algorithms (GA) and three variants of particle swarm optimization (PSO) are applied to this problem. In order to show the capabilities of PSO algorithm in absorber design three different absorber designs are optimized. To alleviate local minima problem, a modified lbest PSO (MLPSO) is employed. Simulation results show that MLPSO searches the space faster and more efficiently, proposing it as a strong tool for designing electromagnetic absorbers.