The author considers effective procedures in the ergodic theory for measure preserving transformations. He proves a version of the ergodic theorem for individual random sequences in the sense of Martin-Löf. The approach is based on A. N. Kolmogorov’s algorithmic framework for probability and information theory.