Background: Adequate assessment of 'normal variation' versus 'abnormal status' is particularly difficult for clinicians working with young children who are under 5 years of age and who present with slow language development. There is therefore clinical motivation to identify possible key difficulties (or 'risk markers') that may distinguish children who are likely to have specific language impairment (SLI) from the variation observed in younger, normally developing children. Aims: The issue of 'risk markers' for SLI was explored. It is well known that the issue of markers is controversial. The view presented here is that a risk marker represents a symptom with no assumption about whether the symptom reflects a single cause or that this symptom alone identifies the disorder. Methods and Procedures: The performance of 32 children with SLI was compared with that of 32 younger normal language-learning children on four potential risk marker tasks: non-word repetition, digit recall, past tense provision and plural marking. Outcomes and Results: The findings suggest that processing markers, particularly non-word repetition, have the potential for indicating SLI risk. In particular, children who fall in the bottom quarter of the normal distribution in non-word repetition (performance below the 25th centile) appear to be at risk of SLI. Conclusions: Although these results need to be interpreted cautiously, the evidence is thought to warrant the use of non-word repetition tasks in clinical practice.