During the last decade new attention has been paid to reading fluency. One reason might be that training studies often have failed to provide growth in reading rate in spite of the fact that the accuracy problem was remediated. Recent research has also indicated that, in languages with a consistent mapping between graphemes and phonemes, automatization of decoding processes are more difficult to reach than aquiring accurate decoding skills. Several studies have found that computer-based flashcard training seems to be a productive way to develop reading fluency. The aims of this thesis are partly to replicate some of the previous studies with time pressured exposure and partly to examine whether the condition with time pressure is more effective than the flashcard condition with fixed exposure duration. In the time pressure condition, the exposure duration was varied as a function of accuracy. This thesis reports one main study and two case studies. The participants in the main study were 17 males and 11 females between the ages of 11 and 17. All of them were poor readers, scoring below the 11th percentile on standardized reading test for their age group. The two participants of the first case study were the poorest readers of all. Their decoding strategies differed greatly. The younger of them, a slow reader in the fourth grade, can be seen as “a speller” with most impairment in the orthographic processing and the older, a quick reader in the fifth grade, can be seen as “a guesser” with most impairment in the phonological processing. The four participants in the second case study were moderately poor readers between the ages of 12 and 14. Each participant practiced the two different conditions of the computer-based flashcard training with spelling response in sessions about twenty minutes long twice to three times a week during three to four months. Teacher-controlled tuition was only given in some pre-sessions. According to the online results, all participants in both of the exposure conditions were capable of keeping the accuracy at a high level, i.e. fluent reading was expected to be developed. A battery of silent and oral reading tests and spelling-to-dictation tests was used in pre-, middle- and posttest. Both accuracy and rate were measured in reading. Except for “the speller” in the first case study, significant improvements were found on all tests for both conditions. The best progress was during the first training period independent of condition. Positive effects were found for both trained and untrained material and for words and pseudo-words. The reading progress has in general been better than the average student usually develops during the same time. The students reported the time pressured condition as being more effective and motivating but no significant difference was found between the two conditions in the pre- to posttests.