BACKGROUND: The shift to test-based management of malaria represents an important departure from established practice under the Integrated Management of Childhood Illnesses (IMCI). The possibility of false results of tests for malaria and co-morbidity, however, make it important that guidelines in IMCI case assessment are still followed. METHODS AND FINDINGS: We conducted a cross-sectional observational study to evaluate current practices in IMCI-based assessment of febrile children in 10 health centres and 5 district hospitals, with follow up of a subset of children to determine day 7-10 post-treatment clinical outcome. Clinical consultation, examination and prescribing practices were recorded using a checklist by trained non-medical observers. The facility case management of 1,983 under-five years old febrile children was observed and 593 followed up at home on days 5-10. The mean number of tasks performed from the 11 tasks expected to be done by the IMCI guidelines was 6 (SD 1.6). More than 6 tasks were performed in only 35% of children and this varied substantially between health facilities (range 3-85%). All 11 tasks were performed in only 1% of children. The most commonly performed tasks were temperature measurement (91%) and weighing (88%). Respiratory rate was checked in only 4% of children presenting with cough or difficulty in breathing. The likelihood of performing "better than average number of tasks" (>6) was higher when the consultation was done by medical assistants than doctors (O.R. = 3.16, 1.02-9.20). The number of tasks performed during assessment did not, however, influence clinical outcome (O.R. = 1.02, 0.83-1.24). CONCLUSION: Facility-tailored interventions are needed to improve adherence to IMCI guidelines incorporating test-based management of malaria. Studies are needed to re-evaluate the continued validity of tasks defined in IMCI case assessment guidelines.