Much of the literature on reading development focuses on measures of early literacy skills (e.g., phonological awareness, phonics, fluency). Elementary educators interested in improving students' skills in these areas can draw on a wealth of research studies. However, many studies of early literacy skills have not addressed comprehension, obviously an important feature of literacy. The authors used a one-parameter Rasch model to examine the relative difficulty of different multiple-choice reading comprehension items assessing students' literal, inferential, and evaluative comprehension of fictional narratives. They also examined the difficulty of questions derived from concepts identified by state content standards as important components of reading comprehension, such as character, prediction, and plot sequence. The findings suggest a curvilinear relationship between literal, inferential, and evaluative comprehension, with literal comprehension being the easiest and inferential and evaluative comprehension more challenging. The findings also indicate that assessment objectives differ on the basis of difficulty. Implications of these findings for practice are discussed. (Contains 3 tables.)