Although proponents of extensive reading (ER) have recommended easy reading material, book difficulty has been poorly defined and operationalized in previous studies. The present study argues for the use of perceived text difficulty for operationalizing book difficulty based on empirical findings (Holster et al., 2017), reading purposes in ER, and the importance of readers' perspectives. A total of 162 Japanese English-as-a-foreign-language university students rated the difficulty of 15 texts excerpted from graded readers (GRs). The data were analyzed by conducting a many-facet Rasch analysis (Linacre, 1989; Rasch, 1960/1980), where a rating scale model (Andrich, 1978) was tested with persons, texts, and graded readers' levels as the facets of measurement. The results revealed that perceived text difficulty could not replicate the stated difficulty level provided by the GR publisher, reinforcing the necessity of examining perceived text difficulty in ER research and practice in the second and foreign language classroom.