The present paper reports on a 5‐year follow‐up of a randomized reading intervention in grade 3 in Sweden. An intervention group (n = 57) received daily training for 12 weeks in phoneme/grapheme mapping, reading comprehension and reading speed, whereas a control group (n = 55) participated in ordinary classroom activities. The main aim was to investigate if there were remaining effects of the intervention on reading‐related skills. Previous analyses showed that the intervention group performed significantly better than the control group on spelling, reading speed, reading comprehension and phoneme awareness at the immediate post‐test with sustained effects 1 year later. Results from the 5‐year follow‐up show that the only significant difference between the intervention (n = 47) and the control group (n = 37) was on word decoding. There was also a significant interaction effect of group assignment and initial word decoding, in the way that the lowest‐performing students benefitted the most from the intervention. Another aim was to examine if the children identified in a screening (n = 2212) as poor readers in grade 2 still performed worse than typical readers. The analyses showed that the typically developing students (n = 66) outperformed the students identified as poor readers in grade 2 on working memory, spelling, reading comprehension and word decoding. Copyright © 2016 John Wiley & Sons, Ltd.