A Bayesian, optimization procedure, in combination with density functional theory calculations, was applied to a combinatorial problem in chemistry. As a specific example, we examined the stable structures of lithium graphite intercalation compoun& (Li-GICs). We found that this approach efficiently identified the stable structure of.stage-I and -II Li-GICs by calculating 4-6% of the full search space. We expect that this approach will be helpful in solving problems in chemistry that can be regarded as a kind of combinatorial problem.