The clinical efficacy of routine admission complete blood count was evaluated in 302 patients admitted to internal medicine wards of a university teaching hospital. Patient medical problems, physical findings and medication history were evaluated by preset criteria to determine the proportion of tests performed for screening and the proportion of test results directly influencing patient management. Of the 282 complete blood counts performed, 80% were ordered routinely with no medical indications (screening tests). An haemoglobin abnormality was found in 16.7% of the patients, leucocyte abnormality in 16.1% and platelet abnormality in 4.6%. However, these results directly influenced patient management in only one case (0.14%). It is concluded that the utility of screening admission complete blood counts in medical inpatients is negligible.