BACKGROUND: WHO now recommends test-based management of malaria across all transmission settings. The accuracy of rapid diagnostic test (RDT) and the outcome of treatment based on the result of tests will influence acceptability of and adherence to the new guidelines. METHOD: We conducted a study at the Kintampo hospital in rural Ghana to evaluate the performance of CareStart, a HRP-2 based RDT, using microscopy as reference. We applied IMCI treatment guidelines, restricted ACT to RDT-positive children and followed-up both RDT-positive (malaria) and RDT-negative (non-malaria) cases over 28 days. RESULTS: 436 children were enrolled in the RDT evaluation and 391 (children with haemoglobin >8.0 gm/dl) were followed-up to assess treatment outcomes. Mean age was 25.4 months (s.d. 14.6). Sensitivity and specificity of the RDT were 100.0% and 73.0% respectively. Over the follow-up period, 32 (18.5%) RDT-negative children converted to positive, with 7 (4.0%) of them presenting with fever. More children in the non-malaria group made unscheduled visits than children in the malaria group (13.3% versus 7.7%) On all scheduled follow-up visits, proportion of children having a temperature higher than that recorded on day 0 was higher in the non-malaria group compared to the malaria group. Reports of unfavourable treatment outcomes by caregivers were higher among the non-malaria group than the malaria group. CONCLUSIONS: The RDT had good sensitivity and specificity. However a minority of children who will not receive ACT based on RDT results may develop clinical malaria within a short period in high transmission settings. This could undermine caregivers' and health workers' confidence in the new guidelines. Improving the quality of management of non-malarial febrile illnesses should be a priority in the era of test-based management of malaria. TRIAL REGISTRATION: ClinicalTrials.gov NCT00832754.