A study to test the validity of the Urdu version of the Strengths and Difficulties Questionnaire (SDQ) was carried out in Pakistan. The study population consisted of 212 children aged 4 to 16 years presenting to psychiatric (case group: n = 72) and paediatric outpatient clinics (comparison group: n = 140). Parents of these children were interviewed using the extended version of the SDQ. The case group was further subjected to diagnostic ICD-10 criteria by child psychiatrists, who were blind to SDQ scores. The Urdu version was able to discriminate between the study groups. The mean difficulties score was higher in the case group (mean = 21.7) than the comparison group (mean = 14.4), the difference (mean difference = 7.3) being statistically significant (p < 0.01, 95% CI = 5.3, 9.3) with an effect size of 1.1. Adequate sensitivity (> or = 61%) was obtained for each type of psychiatric disorder in the case group using the abnormal category of the SDQ symptom scores as a cutoff. A simple 'pencil-and-paper' algorithm combining symptom and impact scores further identified children in need of intervention. A ROC curve analysis showed an area under the curve > 0.70 for all scores. The choice of optimal cutoffs for the Urdu version is discussed.