Background: Clinical services in the UK are increasingly delivering 'consultative' methods of intervention rather than 'direct' intensive input for children with receptive and expressive language difficulties, yet there has been little systematic evaluation of these different intervention models. Aims: To investigate the effectiveness of different models of therapy provision for children with specific language impairment between the ages of 4;00 and 4;06 years. Methods and Procedures: Twenty-four children were selected from a specialist waiting list in the London Borough of Lambeth. They were assessed on a range of verbal and non-verbal skills, and randomly assigned to three different intervention groups. Group 1 received direct intensive speech and language therapy weekly over an 8-month period at a child development centre; Group 2 received a nursery-based model of intervention; and Group 3 received review sessions at their local clinic. Outcome and Results: Statistical analysis before the intervention phase revealed no significant differences in scores between the three groups on a range of clinical and parental measures of language, non-verbal skills, play and behaviour. At the end of the intervention period the Intensive group showed significantly greater improvement than the No Intervention group on all clinical and parental measures, and significantly greater improvement than the Nursery-based group on all clinical and parental measures except for expressive grammar. Conclusions and Implications: The results of this small-scale study demonstrate that intensive direct speech and language therapy delivered by speech and language therapists was a more effective model of intervention for this clinical group with severe speech and language impairment.