Background:  Few field trials exist on the impact of implementing guidelines for the early detection of autism spectrum disorders (ASD). The aims of the present study were to develop and evaluate a clinically relevant integrated early detection programme based on the two‐stage screening approach of Filipek et al. (1999), and to expand the evidence base for this approach. Methods:  The integrated early detection programme encompassed: 1) training relevant professionals to recognise early signs of autism and to use the Early Screening of Autistic Traits Questionnaire (ESAT; Dietz, Swinkels et al., 2006; Swinkels, van Daalen, van Engeland, & Buitelaar, 2006), 2) using a specific referral protocol, and 3) building a multidisciplinary diagnostic team. The programme was evaluated in a controlled study involving children in two regions (N = 2793, range 0–11 years). The main outcome variables were a difference in mean age at ASD diagnosis and a difference in the proportion of children diagnosed before 36 months. Results:  ASD was diagnosed 21 months (95% CI 9.6, 32.4) earlier in the experimental region than in the control region during the follow‐up period, with the mean age at ASD diagnosis decreasing by 19.5 months (95% CI 10.5, 28.5) from baseline in the experimental region. Children from the experimental region were 9.4 times (95% CI 2.1, 41.3) more likely than children from the control region to be diagnosed before age 36 months after correction for baseline measurements. Most of these early diagnosed children had narrowly defined autism with mental retardation. Conclusions:  The integrated early detection programme appears to be clinically relevant and led to the earlier detection of ASD, mainly in children with a low IQ.