Delayed language development is common and has serious sequelae into adulthood in terms of educational, social and emotional development. The objective of this study was to test the hypothesis that a group of language-delayed infants detected in the first year of life by the screen developed in the author's previous study, and receiving intervention at that stage, would not show language impairment at the age at which children are usually referred for speech and language therapy, whereas an untreated control group would do so. A sample (n = 122), of very young language-delayed children was divided into matched experimental and control groups. The mean (range) age of the sample was 10.6 (8-21) months. The experimental group received intervention, and both groups were followed up until they were 3 years of age to determine whether there were differences in language development at that stage. The results showed that at 3 years, 85% of controls showed language delay while only 5% of the experimental group did so. The differences in mean language quotient in experimental and control groups were statistically significant (t = 2.701, p < 0.01). At this stage, 30% of the controls had been referred for speech and language therapy, while none of the experimental group had so been. The intervention was found to take relatively little resource and the low dropout rate and high take-up indicated that it was popular with carers.