In this study, 180 Norwegian fifth‐grade students with a mean age of 10.5 years were administered measures of word recognition skills, strategic text processing, reading motivation and working memory. Six months later, the same students were given three different multiple‐choice reading comprehension measures. Based on three forced‐order hierarchical multiple regression analyses, results indicated that the unique contribution of measured skills and processes to performance varied across comprehension tests. In particular, when the test consisted of a longer passage, contained a larger proportion of inferential questions and was answered without access to relevant text passages, the relative importance of word recognition skills seemed to be reduced while working memory emerged as a relatively strong, unique positive predictor of comprehension performance. These findings have important practical implications for the assessment of reading comprehension.