A variable selection approach for generalized linear mixed models by L1-penalized estimation is provided.