This report investigates the selection of appropriate texts for C-Tests, a modified form of the cloze test, for assessing second language learning. The procedure for textbook readability first involved the administration of different texts to sample groups to determine the C-test difficulty of individual texts. At the same time, a variety of different indices for each text were calculated. The indices were based on the texts' sentence lengths and the specific problems of the native German-speaking students in understanding the texts. The results were validated in subsequent test administrations to learners of German as a second and as a foreign language and with two groups of German and Finnish learners of English as a second language. It is concluded that this method is useful for predicting the difficulty of a text for use in C-Tests, and that the results imply the existence of absolute text difficulty (C-processing difficulty). This method may also serve as a new statistical model for examining the relationship between C-processing difficulty and the learner's position on the language learning continuum. (MSE)