There are several ways to do "stress testing" of the heart, and several ways to see how the heart responds to the stress. The stimuli that can be used are the classic treadmill or bicycle, or drugs such as adenosine, dobutamine, and persantine. The treadmill and bicycle are better at reproducing the effects of physical exercise, but many people cannot perform well on them because they are out of shape, have arthritis, do not have legs, and so forth. The drugs are usually well tolerated, are safe (when used appropriately), and are as reliable as the treadmill and bicycle methods.

The ways to see how the heart responds to stress are the classic electrocardiogram, nuclear methods, and echocardiography. The advantage of the electrocardiogram (ECG) is that it is inexpensive. Unfortunately, it is also the least accurate (see below). There are two nuclear methods: thallium and sestamibi. Thallium is better if there is a good chance that the test will be abnormal, or "positive." Sestamibi is better if the odds are that the test will be negative. Sestamibi is a more powerful isotope and has a shorter half-life, so the pictures are more clear. It does not redistribute, however, so reversible defects (see below) can appear to be fixed. This disadvantage can reportedly be overcome by giving a second injection. Echocardiography gives the most information because it shows the actual motion of each part of the heart at rest and during exercise, but proper interpretation requires the physician to be well-trained in its use.

Interpretation of the stress test has two parts: deciding whether the result of the test is normal or abnormal, and deciding what the result means for the person on whom it was performed. The reason for the second step is that none of the stress testing methods is more than 90-95% accurate. That is, if the nuclear or echocardiographic tests are performed on a group of people who have severe blockages in the coronary arteries, only 90 to 95 out of every 100 of the people will have a "positive" result. Put another way, 5 to 10 of every 100 people with severe blockages, for which bypass surgery is appropriate, will have a normal or "negative" result on the nuclear or echocardiographic tests. Also, if any of the tests is performed on a group of people who have no blockages in the coronary arteries, 5 to 10 of them will have a "positive" result on the nuclear or echocardiographic tests. If the ECG method is used, the numbers are 15 to 20 incorrect results in each group. Mathematically, the sensitivity and specificity of the nuclear and echocardiographic results for three-vessel or left-main coronary artery stenosis are no better than 90-95%. For the ECG method, the sensitivity and specifity are about 80-85%. The figures are about 85% and 75% respectively for less severe (one- and two-vessel) disease.

The mathematical formula for interpreting the stress test is called Bayes theorem. It combines (1) the likelihood of disease based on all available evidence except for the stress test (prior probability), (2) the result of the stress test, and (3) the sensitivity and specificity of the stress test, and yields the likelihood of disease based on all of the available evidence (including the stress test result), or posterior probability. Bayes theorem is hard to apply to individual patients because the prior probability, the sensitivity, and the specificity depend on the risk factors and other diseases that the patient may have, and their exact values are not known for most patients. The theorem is useful to keep in mind as a concept, however, because it shows that there is a big difference between saying that the stress test is abnormal and normal, and saying that the person who performed the test does or does not have coronary artery disease.