Despite the fact that reading is one of the most frequently studied topics in psychology, cognitive science, and educational science, research on text reading has made only little progress in the past 100 years. This is true for the more basic research on reading as a cognitive activity, where materials that resemble natural texts are usually neglected in favor of tasks that employ only individual words or sentences. It is just as true for applied research, where studies either follow the lead of the 'few-words-approach' or focus exclusively on the outcome of reading performance - such as memory or comprehension of a read passage. In this latter case, the reading activity itself that brings about a certain memory or comprehension result remains in a black box. The work presented in this dissertation tries to bridge this gap. In particular, the aim is to expand the work of Wallot and Van Orden (2011a, b) on the application of complexity metrics to evaluate the process of reading in complex text reading tasks. Complexity metrics, which quantify the degree of stability, complexity, and interconnectedness of performance, are employed to expand the findings on reading fluency in self-paced reading by Wallot and Van Orden (2011a, b) to the level of eye-movements during reading. While reading performance during self-paced reading is estimated by the intervals between key-presses that the reader employs to reveal each new word or sentence of a text, eye-movements are considered to be a more fine-grained measure of the reading process (Rayner, 1978). Hence, study 1 of this dissertation is an attempt to replicate the findings of Wallot and Van Orden (2011a, b), where the time evolutions of the performance of more fluent readers showed higher stability and greater commonality compared to less fluent readers, and the stability of and commonality between less fluent readers' performances increased with repeated reading of the same text. Study 2 gathers first evidence for what distinguishes the process of reading of an easy text from the process of reading of a difficult text. While the results of self-paced reading in study 1 do not replicate all of the effects reported by Wallot and Van Orden (2011a, b), they confirm that the performances of fluent readers share more dynamic structure than the performance of less fluent readers and that the used non-linear methods can successfully distinguish between these two reader groups. The results of eye-movement dynamics show effects similar to those of key-press intervals and also replicate the gain in stability of less fluent readers with re-reading. This seems to suggest that eye-movements are indeed a more sensitive measure that can pick out differences in performance that are not picked out in key-press intervals. The results of study 2 suggest that high text difficulty constrains the dynamics of reading measures in key-press intervals and eye-movements alike, since key-press intervals and eye-movements during reading show greater stability during difficult text reading compared to easy text reading. In conjunction, the two studies provide strong evidence for the utility of complexity metrics to quantify reading performance. Furthermore, the studies present evidence that key-press and eye-movement measures of reading seem to be driven by similar organizational principles, following similar dynamics during reading tasks. However, the results of the two studies seem to be contradictory with regard to the role of dynamic stability, which seems to indicate higher reading skill in study 1, but more difficult reading in study 2. Different possible interpretations of these results are discussed. It is concluded that the observed reading performance can be seen as a synergy between text and reader, and that the stability and complexity of this synergy holds information about the overall properties of reading tasks.