Can Science Prove that Prayer Works?

Hector Avalos

Prayer has become a new cottage industry. Within the last five years the New York Times has listed as best- sellers at least a half-dozen books extolling the value of prayer in some form.1 Cover stories have appeared in pop-ular magazines like Newsweek, and television programs such as “Dateline NBC” have devoted entire shows to this sub-ject.2 In particular, physician Larry Dossey in his Prayer Is Good Medicine: How to Reap the Healing Benefits of Prayer (1996) and Healing Words: The Power of Prayer and the Practice of Medicine (1993) has popularized the notion that there is scientific evidence that prayer does work.3

While Dossey sometimes denies that he would impose his spiritual beliefs on his patients, his favoritism toward the supposed efficacy of prayer in the Judeo-Christian traditions is evident.4 Another physician, Randolph C. Byrd, has con-ducted a celebrated study that he believes supports the conclusion that “intercessory prayer to the Judeo-Christian God has a beneficial therapeutic effect” in the population he stud-ied.5

But is there scientific evidence that prayer really works? Can you ever know that a prayer was answered by the Christian god? Does even the notion of prayer make sense? I approach these questions as a professional biblical scholar, as an anthropologist trained in scientific methodologies, and as a former Pentecostal faith-healer. I must conclude that there are fatal flaws with the so-called scientific experiments used by supporters of prayer, and there are even greater philo-sophical and theological problems with verifying scientifi-cally that the Christian god answers prayers.6


Probably the experiment cited most often by advocates of prayer is the one performed by Byrd, a cardiologist at San Francisco General Medical Center. According to his report, he studied 393 patients between August 1982 and May 1983. He divided the group into 192 patients who were prayed for, and 201 who were not prayed for.7 He reported that, among other things, the people who were prayed for were five times less likely to develop pulmonary edema. None required endotracheal intubation, and fewer patients died.

The problem with this and any so-called controlled experiment regarding prayer is that there can be no such thing as a con-trolled experiment concerning prayer. You can never divide people into groups that received prayer and those that did not. The main reason is that there is no way to know that someone did not receive prayer. How would anyone know that some distant relative was not pray-ing for a member of the group that Byrd had identified as having received no prayer? How does one control for prayers said on behalf of all the sick people in the world? How does one assess the degree of faith in patients that are too sick to be interviewed or in the persons performing the prayers? Even Byrd acknowledges these problems and admits that “`pure’ groups were not attained in this study.”8 Since con-trol groups are not possible, such purported scientific exper-iments are not possible.

As noted by Gary P. Posner (FREE INQUIRY, “God in the CCU?” Spring 1990), the empirical results reported by Byrd do not inspire much confidence either. For example, 13 patients (7%) in the prayed-for group died, compared with 17 (8.5%) in the control group. Even Dossey admits that these and other differences between the two groups of Byrd’s patients were statistically insignificant, and concludes: “Do we know any more about the possible effects of prayer from this experiment? I am afraid the answer may be no.”9

To avoid the problems inherent in experiments on human beings, Dossey turns to experiments on nonhuman subjects to support his idea of the efficacy of prayer. Dossey, in fact, sees experiments on bacteria and mice as even more con-vincing because he believes that psychological factors of patients are thereby eliminated. He cites a catalog of experi-ments performed on, among other things, bacteria, yeast, and crystals. Such experiments were published in the American Journal for Psychical Research, the Journal of Para-psychology, Research in Parapsychology, and other journals Dossey considers to have “peer review standards as rigorous as many medical journals.”10

Yet experiments on nonhuman subjects will not help Dossey because these experiments can encounter the same theological and scientific obstacles that plague experiments on human subjects. For example, there are people praying for the well-being of all life on Earth, and so you would not be able to divide bacteria, fungus, mice, or any other living thing into prayed-for and nonprayed-for groups.

None of these experiments (human and nonhuman) have been replicated by those who are generally skeptical of sci-entific studies of prayer. In general, such experiments will probably not inspire confidence until they are at least per-formed by teams of scientists that include both skeptics and supporters of the efficacy of prayer.


But there are worse problems for the prayer hypothesis than a total lack of convincing data. There are profound philo-sophical and theological difficulties in the notion of prayer, especially prayer to the Christian god.

Prayer and the infinite

For Christian believers, answered prayers qualify as a type of miracle. According to Charles Hodge, the famous American fundamentalist theologian: “A miracle, therefore, may be defined to be an event, in the external world, brought about by the immediate efficiency, or simple volition, of God.11 The problem with verifying sci-entifically that miracles as defined above ever occur is that the Christian god is supposed to have infinite characteristics, and we can never know whether a prayer has been answered by a being that is said to be infinite.

Let me explain. One of the infinite characteristics of the Christian god is omnipresence—that is, this being is said to be everywhere in the universe at the same time. The Christian god is also said to be eternal, all-powerful, and all-knowing. Yet, we, as finite human beings, could never know that such an infinite being exists. For example, in order to know that there is a being who is everywhere at the universe at the same time we would have to be everywhere in the universe at the same time. In order to know that there is a being who is eternal, we would have to be eternal. A supposed revelation from a supposedly infinite God will not overcome this problem.12

In order to know that any event we witnessed in the world was caused by a particular being, we first have to know that such a being exists. For example, it would be absurd to say: “I know my prayer was answered by an invisible Martian, but I do not know if invisible Martians exist.” The reason this statement is logically absurd is that it attributes an action to a being not known to exist.

Likewise, in order to know that any event (e.g., an answered prayer or any other supposed extraordinary event) was caused by an infinite being, we first have to know that an infinite being exists. Since we can never know that an infinite being such as the Christian god exists, we can never know that any event we witness was caused by this being. In sum, knowing scientifically that an infinite God answered a prayer is logically impossible.

Pointless prayer

Prayer would be unnecessary if there were an all-knowing, all-good, and all-powerful God. Let’s sup-pose that the most gifted doctor in the world happens to be your friend. This doctor has the ability to cure any sickness known to modern medicine. Let’s also suppose that this doc-tor is living with your family, which includes a six-month-old baby.

Now if this infant were to become violently ill in the pres-ence of this super-doctor, what would you expect from him? If the baby is choking, for example, you would expect him to use techniques that will relieve the baby’s problem. You would not expect him to ask you first if you believed that he could cure your child before he was willing to help the child. You would not expect him to require you to show how much faith you had in him before he would help your child. What you would expect is for this super-doctor to act as soon as he sees the child choking.

Let’s also suppose that this doctor has the ability to pre-vent cancer in all children anywhere in the world even before it occurs. Undoubtedly, you would expect that if he had this ability then he would use it, if he really fits our definition of “good.” But if the doctor has this ability, and does use it, then you would not expect there to be any cases of infantile can-cer in the world. If this super-doctor has this ability, then he should not wait for anyone to ask him to prevent the suffer-ing of children with cancer. We would expect him to act immediately out of pure goodness.

Similarly, an all-good God would not want anyone to suf-fer. An all-knowing God would know who would suffer ahead of time, and an all-powerful God could prevent suffer-ing before it happens. Thus, if there were an all-good, all-knowing, and all-powerful God, then there would be no need for prayer in the first place, especially if the prayer is used to alleviate illnesses or any other type of suffering.

Other gods Even if someone prayed to the Christian god for healing and that person was healed, it would not prove that the healing was done by the Christian god. All religions claim to have answered prayers. For example, according to the Bhagavad-Gita, part of the sacred scriptures of Hinduism, the god Krishna claims that it does not matter which god human beings worship; it is Krishna who answers their prayer.13 Thus, it would not be scientifically possible to show that it is the Christian god who answered a prayer even if such a prayer was answered.

Supernatural ignorance

Even if we saw an extraordinary healing occur (e.g., a severed leg grow back instanta-neously), we would not be able to prove scientifically that it was a supernatural occurrence. To say that something is supernatural is to say that something is not natural. But to say that something is not natural, one would have to be practi-cally omniscient because that would be tantamount to saying that we know all the natural factors that could possibly be responsible for an event, and are claiming to know that none of the factors was responsible. No one has the kind of knowl-edge, and so consequently no one could ever call anything non-natural.

The most we could say about an event whose cause is unknown is that the cause is unknown. As already noted, we would be less justified in attributing an extraordinary event to an infinite being.

Bible problems

Byrd, Dossey, and many other similarly minded scientists usually are uninformed about the Bible and so do not seem to realize that even the Judeo-Christian scrip-tures severely undermine the possibility of controlled scien-tific experiments of prayer. For example, there are many pas-sages that indicate that the Hebrew god will not listen to prayer when he is angry with a whole nation or significant portions of nation’s institutions. Thus, in Isaiah 1:15, the Hebrew god says:

When you stretch out your hands, I will hide my eyes from you; even though you make many prayers, I will not listen; your hands are full of blood.

According to Jeremiah 11:14 the Hebrew god even admon-ishes some not to pray.

As for you, do not pray for this people, or lift up a cry or prayer on their behalf, for I will not listen when they call to me in the time of their trouble.

In the case of nonhuman objects of prayer, consider the reasons for poor plant growth discussed in Haggai 1:9-10:

You have looked for much, and, lo, it came to little; and when you brought it home, 1 blew it away. Why? says the LORD of hosts. Because my house lies in ruins, while all of you hurry off to your own houses. Therefore the heavens above you have withheld the dew, and the earth has withheld its produce.

Thus, no matter how much people prayed in this case, it would not help plant growth until a temple was built.

So how would a so-called controlled study of prayer man-age all of the possible factors (e.g. the building of a temple, anger towards a nation, geographical location of the objects of prayer, etc.) that might lead the so-called Judeo-Christian god to listen to some prayers but not others? Clearly, Dossey, Byrd, and their ideological colleagues do not realize that the many factors mentioned in the Bible for the selective answer of prayers render any thought of a controlled scientific study absolutely meaningless.


For most of my young and adolescent life, I was a faith healer in a Pentecostal tradition. I witnessed what I then thought were resurrections, spontaneous growth of short limbs, cures from cancer, and many other types of diseases. In retrospect, I have learned much about why people believe in answered prayers even when there is evidence to the con-trary or even when it is logically absurd. Every single case of a supposedly answered prayer that I witnessed can be explained by one or more of the following factors: (1) false assumptions, (2) erroneous information, and (3) wishful thinking.

For example, many people with high blood pressure would call me to pray for them when their blood pressure rose. I would come and pray, and afterwards the blood pres-sure would fall. This would be regarded by me and the patient as an answered prayer. Yet most blood pressure fre-quently does rise and fall on its own because our bodies have systems that function like the thermostat in our homes. Many other “sick conditions” also get better on their own because the body has mechanisms to relieve itself (for example, fevers, colds, many types of aches and pains).

Another reason for the widespread belief in divine healing among Christians, especially Pentecostals, is the dynamic of the services in which healings are said to occur. In many instances a great quantity of healings are reported by travel-ing evangelists. Usually the evangelist asks the patient what the problem is.

Many may say, for instance, that they had a “kidney prob-lem” when they have a backache. The evangelist usually does not verify if the patient is indeed suffering from kidney problems and is not usually familiar with the patient’s med-ical history. Yet he might announce that the patient was healed of “kidney problems” to the entire audience. The evangelist also might assume that the persons who approached the altar were healed, and so he may report that multitudes of persons were healed in his previous stop. Indeed, the evangelist rarely performs follow-up examinations. Thus exaggerated numbers of reported healings can multiply rapidly in these environments.

The psychology of the petitioner is also a contributing factor. If the evangelist, for example, asks patients if God has healed them, they are very likely to say “Yes,” even if their symptoms say the opposite.14 The reason is that many patients are embarrassed to say that God has not healed them because this appears to insult God. Many times, patients will say that they have been healed because they really believe that eventually God will heal them, not because their symp-toms have disappeared. Cases where patients say that they have not been healed are attributed to a lack of faith on their part or on the will of God.

Other reports of divine healing are precipitated by the erroneous assumption that doctors cannot be wrong. For example, there are many cases in which a doctor might tell a patient that he or she has an incurable illness from which death will occur within a short period of time. At the end of this period, the patient is not only alive, but much better. This turn of events will usually be attributed to a divine healing. Yet the patient may overlook the fact that the doctor’s diag-nosis may have been wrong in the first place.

The fact is that errors by physicians are more common than most people think, which is reflected by the great number of lawsuits against physicians today. Not all doctors have the same training or experi-ence, or even achieve the same grades in medical school. Again, all these factors make it very simplistic to call a miracle what is more likely the result of human error about a prognosis or a diagnosis.

I, myself, have experienced a mistaken prediction of death. In late 1978, and more dramatically in 1979, I began to develop a disease called Wegener’s Granulomatosis, but it was frequently misdiagnosed by doctors until 1980. The cause of the illness is unknown, but it appears that the body’s immune system begins to attack its own tissues as though they were foreign implants. In 1986, one doctor told me that I would not live past 1988 without major surgery. Although I am not 100% healthy, I am still alive, and the doctor has admitted that he was wrong.

Many improvements in medical conditions are called mir-acles because there is still much to learn about the natural recuperative abilities of the body. For example, even until recently cancer was thought by many to be incurable. Yet today it is known that there are hundreds of types of cancer, and not all of them are incurable.15 Now it is known that the body has many substances that provide greater resistance against cancer in some persons, and in some cases these sub-stances even result in the reversal of cancerous conditions that were previously thought to have been incurable in every case. There are differences in biochemical abilities of per-sons to resist disease. This is why some people can smoke every day of their lives yet live until 90 years of age without major problems, while most people might die at a younger age from smoking. Many researchers argue that stress or other emotional states can affect the ability of the body to fight or recuperate from disease.16

But even if you recovered from a potentially deadly ill-ness in some unexpected manner, you still cannot know if it was an act of God. The most we could say is that the recov-ery was accomplished through an unknown process. Many recuperations that may appear supernaturally miraculous may be due to very natural processes which have not been recognized or studied previously. Indeed, one can draw up a long list of phenomena that were unknown 100 years ago but are deemed perfectly natural today. In fact, most believers in prayer have received conventional medical treatment, and so one cannot eliminate the possibility that it was the medical treatment, not the prayer, that actually had a beneficial effect, even when such an effect might be unexpected.

In all of this we must not forget that there is a dark side to prayer. I’ve seen people die or suffer unnecessarily because they waited too long for their god to answer a prayer. Belief in prayer also opens the door for those who prey on the fragile hopes of vulnerable patients. Dossey, Byrd, and other likeminded “physicians” should focus on helping people manage med-ical problems with real medical solutions, and help people confront reality when there are no medical solutions.



All quotations of the Bible are from the New Revised Standard Version (National Council of Churches of Christ in the United States of America; Nashville: Thomas Nelson, 1989).

  1. Some of these books include: Andrew Weil, Spontaneous Healing (New York: Knopf, 1995); Deepak Chopra, Creating Health: How to Wake Up the Body’s Intelligence (Boston: Houghton Mifflin, 1995).
  2. Most recently on December 24, 1996.
  3. Larry Dossey, Prayer Is Good Medicine: How to Reap the Healing Benefits of Prayer (New York: HarperCollins, 1996).
  4. For example, many arguments (e.g. Healing Words, pp. 68-69) con-cerning the efficacy and benefits of prayer quote Luke 18:1, Romans 12:12, and other passages of the Christian Bible.
  5. Randolph C. Byrd, “Positive Therapeutic Effects of Intercessory Prayer in a Coronary Care Unit Population,” Southern Medical Journal 81 (July 7, 1988): 826-829.
  6. 1 shall focus on discussing the claim that one can show that prayer works because it is being answered by a supernatural being, rather than the relatively more plausible claim that belief in prayer in some instances might produce positive emotional states that can affect a patient’s ability to fight or recuperate from illness.
  7. “Positive Therapeutic Effects,” pp., 826-29.
  8. “Positive Therapeutic Effects,” p. 829.
  9. Healing Words, p. 185.
  10. Healing Words, pp. 211-12.
  11. Charles Hodge, Systematic Theology; 3 vols. (Grand Rapids: Mich.: Eerdmans, 1973 [Reprint]), 1:618.
  12. For example, even if an infinite being appears to us and tell us he is infinite, we would have no means to verify that the being in front of us was infinite. Even 2 Corinthians 11:14 warns Christians that the Devil can disguise himself as an angel of light.
  13. Winthrop Sargeant translates this passage as follows: “He, who, endowed with this faith, desires to propitiate this (God), receives from thence his desires because those desires are decreed by me.” The Bahgavad-Gita (Garden City, N.Y.: Doubleday, 1979) p. 352.
  14. Such behavior has been reported by E. Mansell Pattison, a medical doctor. He studied 71 reports of divine healings in the Seattle, Washington, area in his article “Ideological Support for the Middle Class: Faith Healing and Glossolalia,” In Irving I. Zaretsky and Mark P. Leone, Religious Movements in Contemporary America (Princeton, N.J.: Princeton University Press, 1974), pp. 418-55. According to his analysis, many patients reported healing even when their symptoms had not disappeared.
  15. See John Rennie and Ricki Rusting, “Making Headway Against Cancer,” Scientific American 275 (September 3, 1996): 56-59.
  16. For a review of such studies, see Theodore Melnechuk, “Emotions, Brain, Immunity, and Health: A Review,” In M. Clynes and J. Panksepp, eds. Emotions and Psychopathology (New York: Plenum Press, 1988), pp. 181-247.

Hector Avalos

Hector Avalos is professor of Religious Studies at Iowa State University in Ames, Iowa. Avalos is one of the few openly atheistic biblical scholars in academia. He is the author or editor of nine books, including Fighting Words: The Origins of Religious Violence (Prometheus Books, 2005), The End of Biblical Studies (Prometheus Books, 2007), and Slavery, Abolitionism, and the Ethics of Bliblical Scholarship (Sheffield Phoenix Press, 2011).