We study the relationship between workload and the test ordering behavior of physicians in an operational context where examination can be finished with a test order. We define workload in two forms: the unfinished workload, that is, the work waiting to be completed, and the finished workload, that is, the already completed work since the beginning of the work day. We investigate their effect on the probability of a test order and the number of tests ordered for a patient who receives a test order. Empirical analysis of data from a public research and training hospital shows that these workload measures have different effects: A higher unfinished workload increases the probability of giving a test order, while a higher finished load decreases the number of tests ordered for those patients who receive a test order. The observed effects of the operational factors workload and finished load, which have no relation to the patients' complaints and hence, should have no bearing on the eventual diagnosis, and have significant implications for the quality and cost of health care. We also observe that the control variable, examination time, has a significant negative effect on the likelihood of ordering and the number of tests. We report the results of several robustness tests, which confirm our results.