BACKGROUND: Urinary tract infection in infancy continues to be underdiagnosed, despite its association with renal scarring and thus hypertension, renal failure, and other sequelae. Low ascertainment of urinary tract infections reflects the many difficulties in establishing a diagnosis, some of which could be eliminated by a simple, reliable method for preliminary investigation of children's urine. AIM: To assess the accuracy of a new, simple method for testing urine for nitrite and leucocyte esterase, which could be applied to children in primary care. METHODS: An in vitro study was carried out to compare the results of conventional urine analysis with urine analysis on urine soaked on to panty-liners, and with the laboratory investigation. Two urine analysis stick types were used (Boehringer Mannheim Nephur sticks and Bayer Multistix 8SG) and two brands of panty-liners. Analysis examined evidence of agreement and bias for different methods in addition to sensitivity, specificity, and negative predictive values for urine analysis. RESULTS: Pressing urine analysis test sticks on to panty-liners soaked with urine achieved consistent results compared with the results of conventional dipstick urine analysis. At a prevalence of 21.8%, sensitivity and negative predictive values of urine analysis for laboratory confirmed urinary tract infection were 94% and 98%, respectively, for Boehringer sticks, and 76% and 93%, respectively, for Bayer sticks. At prevalences of 5% and 1% (prevalences that could be expected in primary care) Bayer sticks had negative predictive values of 98.7% and 99.7%, respectively, and Boehringer sticks had values of 99.6% and 99.9%, respectively. CONCLUSIONS: Testing urine on panty-liners is accurate compared with conventional urine analysis. It may be possible to apply this method to testing unwell children presenting in primary care to identify those who require microbiological urine culture to confirm or eliminate a diagnosis of urinary tract infection.