The　problem　of　the　iterative　learning　control　based　on　l2-l∞　filter　is　solved　for　stochastic　discrete-time　systems　with　random　delays.　The　random　delays　are　described　by　using　a　binary　switching　sequence　satisfying　a　conditional　probability　distribution.　Filters　are　obtained　such　that　the　filter　error　systems　are　exponentially　mean-square　stable　and　guarantee　a　prescribed　energy-to-peak　disturbance　attenuation　level　in　term　of　linear　matrix　inequalities　(LMIs).　The　iterative　learning　control　based　on　the　filter　is　designed.　The　simulation　results　show　the　effectiveness　of　the　proposed　design　approach.