The effect of listing fewer laboratory tests on the test request form on test-ordering behavior of a group of 47 Dutch general practitioners was studied. The number of laboratory tests ordered by this experimental group during a 12-month period was recorded. The usual, old standard form was used by a control group of 28 general practitioners in a different region. After having used the old, standard form with 178 tests for 5 months, the experimental group received forms listing only 15 hematological and chemical tests plus several urine and feces tests, and space was allowed for "others." A comparison of the experimental and control periods showed that the number of tests ordered monthly in the experimental group was reduced by 18%. When the usual standard form was re-introduced, the general practitioners quickly returned to their old pattern. Results revealed that fixed and often unsound routine influences the use of additional diagnostic procedures. In addition, limiting and restructuring the test-ordering forms may break the routine but does not essentially modify the rationale of test-ordering behavior.