We investigated the processing of agreement marking and case marking in the comprehension of German relative clauses in 48 seven-year-old monolingual German-speaking children in a picture selection task. We examined the relation between the effectiveness of these different morphological cues and individual memory resources as measured by a digit-span test. Relative clauses were disambiguated either by agreement marking on the embedded verb or by case marking on the embedded determiner phrase. The results show that subject-extracted relative clauses (subject RCs) are more easily comprehended than object-extracted relative clauses (object RCs) and that case disambiguation is more effective than agreement disambiguation. Moreover, we found a relation between individual phonological short-term memory and the effectiveness of the different disambiguating cues in the comprehension of object RCs but not in the comprehension of subject RCs. We found that: (i) children with a low-digit span score have difficulties in the comprehension of case and agreement disambiguated object RCs, (ii) children with a medium-digit span score have difficulties in the comprehension of agreement disambiguated object RCs but not difficulties with case disambiguated object RCs, and (iii) children with a larger digit span score have no difficulties in the comprehension of agreement and case disambiguated object RCs. We explain our results under the "Diagnosis and Repair Model" (Fodor & Inoue, 2000) by arguing that children by age 7 have adult-like competence and processing strategies for these structures and that observable differences depend on individual memory resources as measured by a digit span task. (Contains 9 tables, 3 figures and 2 footnotes.)