Health monitoring techniques that utilize structural materials with the ability to diagnose their own condition, so-called self-diagnosis materials, have been under development. The authors have developed two types of electrically conductive fiber reinforced composite to diagnose cracks in concrete structures: a high sensitivity detection sensor and maximum strain memory sensor. Three points bending tests on pre-notched reinforced concrete beam under the cyclic loading is presented using these two self-diagnosis materials, with attention towards the relationship between crack width of the concrete beam and electric resistance. Moreover, effects of volume fraction of carbon particle on memorizing maximum strain are investigated. It has been proved that both self-diagnosis materials are highly effective to detect the cracks in the concrete. And present strain can be obtained by the proposed fiber reinforced plastic composites. Although volume fraction of carbon particle has significant influence on the characteristics of memorizing maximum strain, maximum strain of the concrete structures can be memorized using the appropriate self-diagnosis materials.