Some reports suggest that there is an increase in the number of children identified as having developmental language impairment (Bercow, 2008). yet resource issues have meant that many speech and language therapy services have compromised provision in some way. Thus, efficient ways of identifying need and prioritizing intervention are required. Despite this, children's natural change is rarely formally considered when intervention planning. The role of age, gender and non-verbal IQ in predicting change has also not always been considered. A longitudinal baseline period is important because it may guide teachers and language therapists towards different therapeutic aims/methods for different groups of children. This study reports on 22 children aged 4-7 identified by teachers as having language difficulties. All children took part in a classroom-based semantic language intervention in two phases, the first led by speech and language therapists (SLTs), followed by a phase led by trained learning support assistants (LSAs). Two Natural Change subgroups were compared: One group showed natural change in language over a six-week period before the intervention ran (n = 11; Baseline Improvers) whilst the other group did not show language change in this non-intervention period (n = 12; Baseline Non-Improvers). The groups were well matched on skills at study entry and differed only on score change without intervention. Language skills were assessed at three subsequent time points: pre-therapy, post-SLT-therapy and post-LSA-therapy. In the group as a whole, significant changes were seen across all language measures except expressive vocabulary. Non-verbal cognitive ability showed no change over time, and did not predict language change. There were important differences in the patterns shown by each subgroup depending on natural change without intervention. Notably, Baseline Non-Improvers changed during intervention more than Baseline Improvers. The results suggest that use of a longitudinal baseline may be useful when planning interventions and deciding where to focus limited resources.