Reading speed is achieved based on automatic word recognition and, together with prosody, constitutes an essential link between word recognition and text comprehension. Despite the relevance of reading speed acquisition for success at school, a high percentage of children growing up in poverty contexts face difficulties in achieving automatic word recognition. Consequently, this paper aims to contribute to the understanding of difficulties in reading speed acquisition in children growing in poverty contexts. Two studies were designed. In the first study, in order to explore the origin of difficulties in developing word reading speed, a comparison of the cognitive profiles of children from low-income backgrounds with and without difficulties in this ability was carried out. In a previous study, norms were obtained for accuracy and speed in a word reading test. Participants were 168 6th grade children from several educational institutions attending children growing up in poverty contexts in Buenos Aires, Argentina. In the present study, the same word reading test was administered to 96 6th grade children. Based on the norms obtained in the previous study, two groups of children were identified: a group who performed at or above the 50th percentile in reading accuracy but below the 30th percentile in reading speed and another group performing at or above the 50th percentile in both measures. The first group was made up of 22 children, and the second one, of 46 children. The remaining 28 children were not included in the study because they performed below the 50th percentile in reading accuracy. Additional tests measuring phonological awareness, rapid naming, verbal memory and word spelling were administered to children in both groups. Between-groups comparisons in these tasks showed that children with speed acquisition difficulties underperformed the other group in the tests tapping phonological awareness, rapid naming and spelling. These results suggest that the children in the group experiencing reading difficulties were still using the phonological route for word recognition. The second study aimed to explore whether a specifically designed educational intervention could enable children with low reading speed from the previous study to increase their reading speed.Both groups of children (with and without reading speed difficulties) were administered two additional reading tests: an experimental test comprising target words which would subsequently be included in the training study for the children with reading speed difficulties; and a reading test of additional words and pseudo words not targeted in the training study, but considered transfer items because they comprised sub lexical units that were included in the target words to be trained during the intervention. These same reading tests were re-administered as a post-test, after the reading intervention for the reading speed difficulties group. The training study aimed to promote reading speed via the acquisition of orthographic representations. The intervention involved two weekly individual sessions lasting 20 minutes each. Each child participated of a total of 15 sessions. Each session included repeated and accelerated reading of lexical units, as well as activities for promoting the analysis of sublexical units included in the target words and also present in the transfer pseudo words from the post-test. The comparison between the pre- and posttest performance of the training group showed a statistically significant increase in reading speed both of trained and transfer words, an increase that was not obtained for the comparison group. This result suggests that during the intervention children were able to develop orthographic representations of the trained lexical units, but also of the sub lexical units that were present in both the target and the transfer words. Educational implications from this study point to the importance of repeated and accelerated reading for increasing speed, a critical reading ability.