Psychodynamic Psychotherapy: Criticism and What Is True About It

I am a psychologist and my work is depth psychology-based. Today, that is psychological counseling by video: depth psychology on screen. My responsibility includes both the strengths of the approach and its limits.

The evidence from studies is inconsistent. For depression, short-term psychodynamic and behavioral therapies are, on average, similarly effective; for anxiety disorders, the guideline primarily recommends cognitive behavioral therapy (CBT). For obsessive-compulsive disorder, there has been a first positive randomized trial of short-term psychodynamic therapy since 2026; however, CBT with exposure is the best-studied approach and continues to be recommended as the first-line treatment. For post-traumatic stress disorder (PTSD), trauma-focused approaches shall be offered.

This text deals with depth psychology-based psychotherapy. Together with analytical psychotherapy, it is one of the psychoanalytically grounded approaches included in the Psychotherapy Guidelines, which govern treatment in Germany. Nevertheless, they are two distinct forms of treatment: For adults, individual therapy is limited to a maximum of 100 sessions of depth psychology-based treatment, but up to 300 sessions of analytical treatment.

Likewise, the international research term “psychodynamic therapy” does not correspond exactly to treatment under the German Psychotherapy Guidelines. Many studies examine manualized short-term therapies consisting of 20 to 40 sessions. Their results therefore cannot be readily applied to a treatment of 60 to 100 sessions.

Guidelines summarize the research and translate it into treatment recommendations. The quality of evidence and the strength of a recommendation are not the same thing. Grade A (“shall”) indicates a strong recommendation, while Grade B (“should”) indicates a recommendation. In addition to the available studies, factors such as benefits, risks, clinical experience, and patient preferences are taken into account.

Randomized controlled trials are particularly important for assessing efficacy. Random assignment reduces systematic differences between groups, but does not eliminate every possible source of bias.

Depression. The National Disease Management Guideline for Unipolar Depression, Version 3.2 from July 2023, does not favor one approach over another: “To date, there is no evidence on the differential indication of specific approaches for specific patients with depressive disorders.” This is not a free pass, since the guideline itself adds the caveat: “However, not every approach is equally suitable for every patient,” and a wrong choice could also lead to a worsening. A lack of evidence does not mean that the choice does not matter. According to Recommendation 4-27, the choice belongs in the consultation or the trial sessions.

Anxiety. For anxiety disorders, CBT is the firm first choice: first-line treatment with the highest grade of recommendation. That is what it says in the S3 guideline on the treatment of anxiety disorders, Version 2.0 dated April 6, 2021: “Cognitive behavioral therapy (CBT) shall be offered (Ia/A).” Psychodynamic treatment follows, with grade of recommendation B and subject to one condition: it should be offered “if CBT has not proved effective, is not available, or if the informed patient has a preference for it.” This applies, with identical wording, to panic disorder and agoraphobia, generalized anxiety disorder, and social phobia. Regarding generalized anxiety disorder, the guideline is clear: “Psychodynamic therapy appears to be inferior to CBT.” The guideline’s own addendum belongs in the same place: “The available studies were conducted over up to 30 hours.”

OCD. For obsessive-compulsive disorders, this very evidence is lacking. Statement 4-21 of the S3 guideline on obsessive-compulsive disorder, Version 2.0 dated June 30, 2022, states: “There is no evidence from randomized controlled trials for the effectiveness of these approaches.” Recommendation 4-1, Grade A: “For obsessive-compulsive disorder, disorder-specific cognitive behavioral therapy (CBT), including exposure, shall be offered as first-line psychotherapy.”

The first-line treatment in accordance with the guidelines therefore remains disorder-specific CBT, including exposure. I have also treated people with obsessive-compulsive disorder using psychodynamic approaches, but I have never presented this approach as being equally well-supported. Understanding individual fears, conflicts, and relationship patterns can be helpful; however, it does not replace transparent information about the more well-established treatment of obsessive-compulsive symptoms.

In the standard textbook of my approach, Michael Ermann cites, in addition to the early developmental disorder, a genetic predisposition and involvement of the frontal lobe, basal ganglia, and limbic system. He also writes that obsessive-compulsive disorders are chronic, do not resolve spontaneously, and are extremely difficult to treat.

PTSD. The S3 guideline on post-traumatic stress disorder, in the version of February 27, 2026, recommends, with Grade A, a trauma-focused approach in an individual setting and lists five interventions with solid evidence, including Eye Movement Desensitization and Reprocessing (EMDR). Psychodynamic approaches are mentioned only in the research recommendations and among the approaches whose effectiveness has not been sufficiently tested. However, many practitioners across all disciplines have additional training in trauma therapy approaches.

On average, it is about as effective as established approaches: that is the key finding. A meta-analysis by Steinert and colleagues summarized 23 randomized studies involving 2,751 participants with various disorders. In these studies, manualized psychodynamic therapies were statistically equivalent to established comparison treatments in terms of the target symptoms. The meta-analysis by Smith and Hewitt confirmed this finding specifically for depression, immediately after the end of treatment.

The Cochrane Review by Abbass et al. examined 33 studies on short-term psychodynamic therapy. The results suggested an effect, but because the studies were small and varied widely, they should be interpreted with caution.

The blanket assertion that this approach is always just as effective as behavioral therapy is not supported by the available data. A more cautious statement is accurate: For some disorders, manualized psychodynamic therapies achieve, on average, similar improvements in symptoms as established treatments. For other disorders, longer courses of treatment, and specific patient groups, the evidence remains significantly less certain.

Equivalence in symptoms is supported by evidence. None of the major reviews claims superiority, in either direction. The body of research behind this is small. Two reviews compare my approach directly with others, one compares short-term therapy with a waiting list and standard treatment, and one summarizes the entire field. All four measure symptoms, most of them in manualized short-term therapies. What critics make of them depends on this, and this is where the response of the proponents of my approach begins: with the yardstick and the duration.

An article in the Deutsches Ärzteblatt from 2006, titled “Therapeutic Methods: The Denigration of Depth Psychology,” shows how old and how heated this dispute is. The author criticized behavioral therapy for wrongly portraying itself as superior and claimed that his own approach was supported by sufficient studies in all areas of indication. This claim goes beyond the evidence, both then and now.

In 2017, Steinert and colleagues published a review in the American Journal of Psychiatry in which they formally assessed 23 randomized trials involving 2,751 patients for equivalence, using a predetermined threshold of g = 0.25. At the end of treatment, the difference was g = -0.153, which was below the threshold. The method is noteworthy: representatives from both sides sat together on the working group. The study did not go unchallenged. An objection by Ioana Cristea, Pim Cuijpers, and Florian Naudet, titled “Limited Supporting Evidence and Clinical Relevance,” appeared in the same journal.

The independent cross-check came in 2024 from Smith and Hewitt at the University of British Columbia. In nine studies, they directly compared manualized psychodynamic therapy with manualized behavioral therapy for depression and arrived at g = -0.11 at the end of treatment, formally equivalent. At follow-up within one year, the finding was neither equivalent nor different; there, the available data are insufficient for any conclusion.

The Cochrane Review on short-term psychodynamic therapy by Abbass and colleagues, in its 2014 version, includes 33 studies with 2,173 participants and is based on a database search through July 2012. The authors urge caution because of limited data and large differences between the studies.

A study by the same research group provides the framework: Leichsenring and colleagues analyzed 102 meta-analyses with 650,514 patients in World Psychiatry in 2022. Talk therapy and medication are on par there (0.34 versus 0.36); in a direct comparison, the difference is 0.11. The authors speak of a ceiling effect in treatment research. The Institute for Quality and Efficiency in Health Care (IQWiG) states on depression: “It cannot currently be said with certainty which approaches are most effective in which situations.”

A gap remains, and it is a significant one. Measurements were made on manualized short-term therapies, in Abbass’s review on no more than 40 sessions. The German form under the Psychotherapy Guidelines, at 60 to 100 sessions, is not represented in them. In my experience, the most lasting and profound changes usually occur in the last third of the treatment, once the core issues have been worked through repeatedly.

This accusation is as old as the approach itself, and Freud himself wrote it down. In “Constructions in Analysis” from 1937, he quotes a researcher he held in high regard: “When we present our interpretations to a patient, we deal with him according to the notorious principle: ‘Heads I win, tails you lose.’” If the patient agrees, the interpretation counts as confirmed; if he disagrees, the disagreement counts as resistance and confirms it as well.

Karl Popper therefore described psychoanalytic explanations as unfalsifiable. He was not saying, however, that every one of Freud’s observations was wrong. His objection was directed against claims for which no conceivable counterargument is permitted. Adolf Grünbaum, on the other hand, considered many psychoanalytic assumptions to be testable in principle, but doubted that an improvement alone confirms their correctness: it could also result from other factors.

In his textbook, Ermann writes that the psychodynamic hypothesis can “at first be nothing more than a conjecture” and must be “tested during treatment and revised if necessary.” And whether interpretations have any effect at all has since been tested using random assignment. In 2008, Høglend and colleagues treated one hundred patients for one year, divided into two groups by lot. One group received psychodynamic therapy with transference interpretations, that is, with interpretations of what takes place between patient and therapist; the other received the same therapy without these interpretations. Both groups improved clearly, and across all one hundred patients, both forms did equally well. Only when the researchers split the groups by relationship history did the difference appear. Patients who had experienced difficult relationships throughout their lives made greater progress with transference interpretations than without them, and this advantage persisted until the follow-up examination three years after the end of treatment.

The guideline question and the long-term question are two separate questions, and for the second one, there is data that does not count according to the rules of the first. In the Helsinki study (Knekt et al. 2008), patients were randomly assigned to one long-term and two short-term therapies, none of which was behavioral therapy. In the first year, the short-term therapies were ahead; after three years, the long-term therapy was ahead. In 2023, the German long-term study by Beutel and colleagues compared long-term analytical therapy and long-term behavioral therapy over five years in chronic depression. In terms of symptoms, both were on par; in terms of change in personality structure, that is, precisely the structures Ermann means, analytical treatment was ahead. These studies do not show any superiority in terms of symptoms.

The fact that psychotherapy has side effects is no trivial matter. These include temporary worsening of symptoms, relationship conflicts, dependence on treatment, and emotional strain resulting from confronting difficult experiences.

It is not possible to reduce the frequency of side effects associated with depth psychology-based treatment to a reliable general figure. Definitions, survey methods, and the patient groups studied vary widely.

In the study by Balder and colleagues, clinicians reported at least one side effect in 33.2 percent of 276 cases; there was no statistically significant difference between psychodynamic therapy and CBT. The severity of the illness, however, was associated with the number of reported side effects. In a differently designed survey by Gerke and colleagues, 15.2 percent of former outpatients reported at least one side effect. The figures are not directly comparable.

Damage caused by errors, breaches of boundaries, or inappropriate treatments must be distinguished from side effects resulting from a properly administered treatment.

Above all: A deterioration is not a necessary part of successful psychotherapy and no proof that “something is starting to move.” Temporary stress may occur. If a deterioration persists or becomes severe, the course of action must be reviewed promptly and, if necessary, changed.

Things often get worse before they get better. If you look behind the scenes, you’ll feel it more at first. Most people, however, see this as progress and say so: “I’m feeling worse now than before, and I’m glad I’m facing this head-on.” In the process, many come to realize that the source of their distress has been there the whole time and has been affecting them all along.

Pursuant to Section 630e of the German Civil Code (BGB), patients must be provided with clear and understandable information regarding significant risks, prospects for success, and substantially different treatment options.

Some of the objections are valid. The key theoretical concepts are difficult to operationalize, interpretations can become circular, and the research does not adequately reflect long-term treatment in Germany.

What hits home is the question of testability. Section 16a, paragraph 1, of the Psychotherapy Guidelines directs the approach toward “the unconscious psychodynamics of currently active neurotic conflicts and structural disorders, taking into account transference, countertransference, and resistance.” These are constructs whose presence is difficult to substantiate or refute in an individual case. But it is just as pointless as questioning the spiritual experiences of meditators. It is a question of measurability, which is why psychodynamically based approaches have a hard time in modern science.

Effectiveness can be tested, and it has been tested. For depression and some anxiety disorders, there are randomized trials and meta-analyses. For PTSD, the guidelines clearly favor trauma-focused approaches. For obsessive-compulsive disorder, a first positive trial has been available since 2026, while CBT with exposure has been studied much more extensively. And at present, this is always the case: behavioral therapy approaches are better studied because they are easier to study. This is one reason why psychodynamic approaches are rarely found in research. Imagine you want to be taken seriously as a student, a research assistant at a university, or a professor. In that case, you’d be better off focusing on research areas that can be easily operationalized rather than some abstract constructs about which you know very little, precisely because they are hardly taught at universities anymore.