Avian brood parasites exploit parental care of their hosts. Acceptance of parasitism usually results in a considerable loss of host reproductive success. This cost of accepting parasitism should select for host defense against brood parasitism. In fact, empirical studies have demonstrated that some hosts have established an ability to recognize and reject parasitic egg in response to parasitism, but at the same time that not all hosts show such defense against parasitism, a puzzle that cumbers many ornithologists and evolutionary biologists for years. It has been suggested that the host defense entails some costs and that the cost-benefit balance of acceptance and rejection is the key to understand imperfect host defense. Recent study also has pointed out the importance of intra- and inter-clutch variation on the establishment of host defense. In this study we consider the effect of these variation in egg appearance on the cost-benefit balance. We also consider "conditional rejection" that hosts memorize and learn clutch content in the 1st breeding and behave as rejecters in later breeding attempts based on the learnt experience. We compare three host strategies, naive acceptance, unconditional rejection without learning, and conditional rejection with learning and analyze under what conditions each of these confers highest fitness.