Reading is used in everyday life for obtaining information from various sources and for enjoyment, but also for gaining knowledge. Human brains are able to mentally interpret graphic forms they see on a page or on a screen, and different reading skills and strategies are applied to meet a range of communication needs. Although often considered a passive activity of receiving messages, the reading construct can be explained as integration of many component skills (Grabe, 2014). In other words, it is a set of active mental processing required to decode the written message. Reading in everyday life differs from reading practiced during classes, especially in foreign language classes, and the differences are visible in the materials used and in the very purpose of reading. The context of English as Foreign/Second Language recognises reading ability, especially reading in English, as an essential skill in a range of secondary and higher education settings. To date, many theories of reading have emerged, and the activity of reading itself has proven to be one of the most important during education. Meaning of a text is formed in the process of interaction between a text and a reader. To be able to comprehend some text, a number of things have to be included, such as abilities to recognize words fast and efficiently, develop and use a large recognition vocabulary, process sentences to build comprehension, use various strategic processes and underlying cognitive skills, interpret meaning in relation to reader’s background knowledge, interpret and evaluate texts in line with reading goals and purposes, and process texts fluently over a period of time (Grabe, 2014). Reading is an even more complex construct when a foreign language is involved due to the interaction of two or more language systems. The research of reading can go in two directions: researching reading as a process or examining the product of that process (Alderson, 2002) and both directions certainly have its challenges. The introduction of the State Matura (SM) as a compulsory exam for further education at higher education institutions additionally emphasized the importance of reading in other languages, especially English. This paper investigates the types of reading tasks used in reading tests, tasks in relation to their templates, i.e., their textual features, especially cohesion. The starting point lies in the modern theoretical knowledge about foreign language reading as multilayered data processing which provides insight into the importance of proper selection of tasks to check certain components of reading. This paper deals with reading in the context of knowledge testing in the English language exams within the SM. Reading is explored in the evaluation process with respect to the textual features and types of tasks used to test reading comprehension in standardized SM exams, where a number of challenges are possible. Challenges can refer to the data itself and their interpretation, or to tasks which examine the construct of reading. With data, we need to keep in mind how much they really encompass the concepts we want to examine as well as the possibility of their misinterpretation. The tasks (methods) of data collection can disable or encourage the use of certain language forms, which does not necessarily indicate the student's lack of knowledge. The familiarity of testing methods can also greatly influence the obtained results. Coding and analysis of data, together with the reliability of the given process should also be taken into account. Therefore, in this paper, the emphasis is on the types of tasks that test reading, but also on the data collected throughout standardized testing. The following paragraphs outline the structure of this paper. After the introductory part, the second chapter gives an overview of reading as a process by describing internal, external and interactive models of reading, surveys various reading techniques, and finally elaborates foreign language reading. The third chapter builds on reading in a foreign language by clarifying the concepts of language transfer and language threshold at different levels and outlining the relevant research in this area. Reading skills are developed in the mother tongue and, according to some authors, transferred to another language (Cummins, 1979; Koda, 2012). Back in 1984, Alderson questioned whether the problem of text comprehension was the result of insufficiently developed reading skills or insufficient knowledge of the target language, and the notion of language transfer is logically related (Odlin, 1989, 2002). Cummins (1979, 2000) attempts to clarify this problem with his language interdependence hypothesis and the hypothesis on the level of language threshold required for possible transfer of skills. The fourth chapter discusses the text, its understanding and textual features in more detail, while the following chapter provides an analysis of basic documents relevant to the teaching process in the Republic of Croatia: the curricula, the Croatian National Curriculum Framework and the Common European Framework of Reference for Languages. Qualitative analysis indicates how text and reading are presented in the mentioned documents, and what possible influence they have in language teaching. The sixth chapter then explains the process of testing and evaluating language skills with a special focus on reading. With the introduction of external evaluation at the end of secondary education in many countries, including the Republic of Croatia, a greater emphasis is placed on reading examinations in the context of standardized exams, which are also considered high-risk exams due to their impact on all stakeholders. Finally, the characteristics of the various types of tasks used to evaluate reading in the SM exams in English are explained. The research part of this paper is organized to firstly define the research context, and then to describe the higher-level reading exams since the results of the relevant SM exams from the school year 2010/2011 to 2016/2017 were used as the research corpus. The aim of the research is to determine the influence of various types of tasks and textual features on reading comprehension. The analyses try to identify the factors that lead to poorer achievements in certain types of tasks, as well as the impact of textual features on text comprehension. The main goal of the research is, therefore, to gain insight into the importance of proper selection of tasks and text templates for evaluating certain reading components. To achieve this goal, the following research questions were stated: 1) which type of task causes the least and the most difficulties in text comprehension, 2) do gap-fill tasks test linguistic knowledge or text comprehension, 3) how do textual features affect reading comprehension. The results show that gap-fill tasks are the most demanding for the participants. Also, in solving them there is a visible level of reading comprehension, but also a lack of language knowledge manifested in semantic, morphosyntactic and orthographic errors. The Coh-Metrix tool was used to analyse 35 text templates and identify textual features important for readability, such as the appearance of pronouns, the number of modifiers in noun phrases, the familiarity of lexemes, hypernyms in verbs and nouns, but also dense anaphoric and cataphoric references and discourse markers. The results indicate the importance of choosing the right types of tasks for testing foreign language reading, as well as the inseparability of language knowledge from reading skill, and the great influence of textual features on readability and text comprehension. The analyses try to prove the necessity of harmonizing the characteristics of text templates used in standardized tests, and to determine the inseparability of language knowledge, reading skills, types of tasks and textual features. The obtained results emphasize the importance of the text and tasks in the process of reading evaluation, which is in contrast to the studies focused on the reader and his/her characteristics. The knowledge gained from this research can be used by both teachers and examiners to draw more attention to textual features, especially readability and cohesive devices, but also to carefully choose tasks in relation to the reading segment they want to evaluate. Besides, the findings could be used to help support more effective teaching and to improve students’ reading abilities.