In comprehending a meaning of a sentence, the sentence processor has to determine the relationship between case-marked noun phrases and verbs. Several studies claim that the sentence processor performs that decision process before the input of the verb by using some available information such as case markers. In Japanese, two kinds of case markers are specified for each verb which case marker is assigned to an object NP -- one is a dative case marker (-ni), and the other is an accusative case marker (-o). This paper investigates the effects of the case marker information when the sentence processor predicts following elements to appear. Results of ERP experiments show two findings: (1) the dative marker and the accusative marker make the sentence processor predict different elements in each; an accusative-marked NP and verb are predicted by former and only verb is predicted by latter, (2) only dative or accusative case marker affects the prediction of following input, independent of a nominative case marker (-ga).