To evaluate antibiotic use among patients presenting to primary healthcare facilities with febrile illness in Zambia. We analysed data from a 2011 nationwide cross-sectional health facility survey of routine malaria case management in Zambia. Patient consultation observation and medical record charts were used to calculate the proportion of febrile patients who were prescribed antibiotics, stratified by symptoms, health workers' diagnosis and malaria test results. Logistic regression was used to identify factors affecting antibiotic prescribing behaviour. Of 872 patients presenting with fever, 651 (74.6%) were tested for malaria. Among those tested, 608 (93.4%) had analysable results; 230 (37.8%) had positive results. Antibiotics were prescribed to 69/230 (30.0%), 247/378 (65.3%) and 132/221 (59.7%) of those who tested positive, negative and those 'not tested', respectively. Furthermore, antibiotics were prescribed to 36/59 (61.0%) and 242/322 (75.1%) of those diagnosed with diarrhoea and upper respiratory tract infection (URTI), respectively. Among patients prescribed any antibiotic, concurrent antimalarial prescribing occurred in 66/69 (95.6%), 32/247 (12.9%) and 19/132 (14.4%) for those with positive results, negative results and 'not tested', respectively. Respiratory symptoms, diagnosis of URTI, malaria or skin disease and level of health care in the health delivery system were associated with antibiotic prescribing. Testing positive for malaria or receiving a malaria diagnosis was associated with reduced antibiotic prescribing, while testing negative, not being tested or a diagnosis of URTI resulted in higher rates of antibiotic prescribing. There is a need for improving diagnostic capacity for non-malaria causes of febrile illness at healthcare delivery points and limiting antibiotic use to patients with definite bacterial infections.