Three-Factor Eating Questionnaire – Revised 18 (TFEQ-R18)

Abstract

The Three-Factor Eating Questionnaire – Revised 18 (TFEQ-R18) is a psychometrically refined self-report instrument designed to evaluate three distinct cognitive and behavioral dimensions of eating patterns: Cognitive Restraint (CR), Uncontrolled Eating (UE), and Emotional Eating (EE). Developed by Jan Karlsson, Lars-Olof Persson, Lars Sjöström, and Marianne Sullivan in 2000 through an extensive structural evaluation of Albert J. Stunkard and Samuel Messick’s original 51-item Three-Factor Eating Questionnaire (TFEQ), the 18-item revision resolves historic psychometric limitations, including unstable factor structures, item redundancy, and cross-loadings. The instrument comprises 18 items scored primarily on a 4-point Likert-type response continuum (with a single 8-point global rating item recoded to a 4-point scale), which are mapped to three robust subscales: Cognitive Restraint (6 items), assessing conscious dietary restriction aimed at weight control; Uncontrolled Eating (9 items), capturing episodes of loss of control over food intake triggered by external cues or perceived physiological hunger; and Emotional Eating (3 items), measuring the tendency to overeat in response to negative affective states such as anxiety, sadness, and loneliness. Subscale raw scores are routinely transformed to a standardized 0–100 scale to facilitate clinical interpretation and comparative research. Psychometric evaluations across diverse non-clinical, overweight, obese, and post-bariatric surgery populations demonstrate that the TFEQ-R18 exhibits high internal consistency (Cronbach’s alpha typically ranging between 0.75 and 0.91 across dimensions), superior factorial stability verified via confirmatory factor analysis, and solid criterion-related and construct validity. By offering an abbreviated, statistically sound assessment of maladaptive and regulated eating behaviors, the TFEQ-R18 serves as an essential measurement tool in obesity medicine, clinical nutrition, bariatric psychology, and behavioral intervention research.

Keywords

Three-Factor Eating Questionnaire, TFEQ-R18, Cognitive Restraint, Uncontrolled Eating, Emotional Eating, eating behavior, psychometrics, dietary restraint, obesity assessment, bariatric surgery, food intake regulation

Authors

The Three-Factor Eating Questionnaire – Revised 18 was developed and validated by an interdisciplinary team of clinical researchers, psychologists, and medical scientists affiliated with the Health Care Research Unit, Department of Medicine, and the Department of Primary Health Care at Sahlgrenska University Hospital and Göteborg University, Gothenburg, Sweden:

• Jan Karlsson, Ph.D. — Health Care Research Unit, Department of Medicine, Sahlgrenska University Hospital, Göteborg University, Gothenburg, Sweden. Specializes in behavioral medicine, health-related quality of life, and psychometric instrument development.
• Lars-Olof Persson, Ph.D. — Institute of Health Care Sciences and Department of Primary Health Care, Sahlgrenska Academy at Göteborg University, Gothenburg, Sweden. Expert in clinical psychology, psychometrics, and stress and coping mechanisms.
• Lars Sjöström, M.D., Ph.D. — Obesity Unit, Department of Body Composition and Metabolism, Sahlgrenska University Hospital, Göteborg University, Gothenburg, Sweden. Principal Investigator of the seminal Swedish Obese Subjects (SOS) study and pioneer in obesity and bariatric surgery research.
• Marianne Sullivan, Ph.D. — Health Care Research Unit, Department of Medicine, Sahlgrenska University Hospital, Göteborg University, Gothenburg, Sweden. Renowned scholar in health status assessment, patient-reported outcome measures (PROMs), and behavioral epidemiology.

Purpose

Eating behavior is a multifaceted behavioral domain governed by intricate neurobiological, homeostatic, psychological, and environmental determinants. The original 51-item Three-Factor Eating Questionnaire (TFEQ), formulated by Albert J. Stunkard and Samuel Messick in 1985, represented a paradigm shift in characterizing eating phenotypes by delineating cognitive restraint, disinhibition, and hunger. However, methodological inquiries over subsequent decades exposed notable psychometric vulnerabilities in the original 51-item instrument, including dichotomous true/false formats that restricted response variance, unstable factorial replication across clinical and non-clinical cohorts, and excessive respondent burden. To address these vulnerabilities, Karlsson and colleagues (2000) conducted rigorous psychometric analyses within the prospective Swedish Obese Subjects (SOS) study cohort, purposefully designing the TFEQ-R18 to provide a parsimonious, robust, and structurally invariant measure of eating styles.

The primary clinical and empirical purpose of the TFEQ-R18 is to quantify three cardinal dimensions of non-homeostatic eating: conscious efforts to restrict dietary intake (Cognitive Restraint), subjective susceptibility to overeating induced by physiological or environmental triggers (Uncontrolled Eating), and food consumption provoked by negative emotional distress (Emotional Eating). In clinical contexts, such as bariatric surgery evaluations and medical weight-management clinics, the instrument serves as an indispensable baseline assessment and longitudinal monitoring instrument. It enables multidisciplinary healthcare teams to identify maladaptive eating phenotypes that may impair weight loss trajectories, precipitate weight regain, or predispose individuals to subclinical or clinical eating disorders such as binge eating disorder (BED).

In behavioral and epidemiological research, the TFEQ-R18 enables investigators to model eating patterns against physiological biomarkers, genetic profiles, body mass index (BMI), body composition changes, and nutritional regimens. Its brevity markedly reduces respondent fatigue, making it exceptionally well-suited for large-scale epidemiological investigations, clinical trials examining anti-obesity pharmacotherapy, and behavioral lifestyle modifications where repeated-measures testing is essential.

Psychological Construct

The TFEQ-R18 operationalizes eating behavior across three foundational constructs, each capturing a distinct neurocognitive and behavioral manifestation of dietary self-regulation and dysregulation:

1. Cognitive Restraint (CR)

Cognitive Restraint describes the conscious, cognitive monitoring and deliberate effort to curtail, restrict, or avoid food consumption for the explicit purpose of controlling body weight or modifying body shape. Unlike intuitive or homeostatically guided eating—where food intake is dictated by internal hunger and satiety cues—individuals characterized by high cognitive restraint rely on mentally imposed rules, caloric budgeting, portion rationing, and food group proscriptions. In the TFEQ-R18, this dimension is represented by 6 items (e.g., deliberately taking small helpings, consciously holding back at meals, and avoiding foods perceived to be fattening). While moderate levels of cognitive restraint can facilitate therapeutic adherence to energy-restricted diets, extreme or rigid restraint often engenders cognitive fatigue and increases psychological vulnerability to counter-regulatory overeating or binge episodes when self-regulatory resources are depleted.

2. Uncontrolled Eating (UE)

Uncontrolled Eating denotes a subjective sense of impaired control over food consumption, characterized by frequent episodes of overeating, ravenous appetitive responsiveness, and an inability to resist enticing culinary stimuli. This construct integrates both external eating (reactivity to the sight, smell, or social context of food) and intense homeostatic hunger sensations that feel insatiable (e.g., feelings that one’s stomach is a ‘bottomless pit’). Represented by 9 items in the instrument, Uncontrolled Eating reflects heightened reward sensitivity and compromised inhibitory control in the presence of hyperpalatable foods. Individuals with elevated scores on this subscale report persistent difficulty ceasing consumption once initiated and an enduring susceptibility to overeat even after recently completing a meal.

3. Emotional Eating (EE)

Emotional Eating captures the behavioral tendency to consume food in direct response to negative affective states and psychological distress, rather than physiological hunger. Comprising 3 items, this construct assesses hyperphagic reactions to dysphoria, loneliness, and anxiety (e.g., consoling oneself through eating, overeating when feeling blue, or turning to food when experiencing acute stress). Emotional eating functions as a maladaptive emotion regulation strategy; palatable, energy-dense foods activate central dopaminergic and opioid pathways, temporarily blunting negative affect at the expense of metabolic dysregulation. Clinically, emotional eating serves as a vital prognostic indicator in metabolic and weight-loss interventions, often mediating the relationship between psychological stress, depressive symptoms, and adiposity.

Theoretical Framework

The theoretical architecture of the TFEQ-R18 is anchored in several foundational models within behavioral medicine, clinical psychology, and neurobiology, primarily drawing upon Restraint Theory, the Externality Theory of eating behavior, and affect-regulation models of appetite.

Restraint Theory, originally posited by C. Peter Herman and Janet Polivy (1975), forms the central intellectual foundation for the Cognitive Restraint subscale. Herman and Polivy postulated that chronic cognitive restriction of food intake fundamentally alters physiological hunger-satiety set points, replacing somatic homeostatic feedback with cognitive control. Under this paradigm, when cognitive control is undermined—by emotional distress, alcohol intoxication, or the forced consumption of a perceived high-calorie ‘preload’—restrained eaters experience the ‘what-the-hell effect’ (counter-regulation), resulting in pronounced disinhibition and acute overeating. The TFEQ-R18 refines this conceptualization by isolating deliberate restraint from its dysregulated consequences, allowing researchers to study restraint as an independent behavioral mechanism.

The Uncontrolled Eating dimension aligns closely with Stanley Schachter’s (1968) Externality Theory, which suggested that individuals prone to obesity are disproportionately responsive to external, salient food-related environmental cues (e.g., olfactory, visual, social contexts) rather than internal visceral sensations of satiety. Contemporary cognitive neuroscience expands upon this framework via Dual-Process Models of self-regulation: eating behavior is conceptualized as an ongoing conflict between an impulsive, cue-driven reward system (subserved by the striatum and amygdala) and a reflective, goal-directed executive control system (subserved by the prefrontal cortex). High scores on Uncontrolled Eating signify a hyper-reactive appetitive reward system coupled with attenuated executive restraint.

Finally, the Emotional Eating construct is rooted in the Psychosomatic Theory of obesity (Bruch, 1961; Kaplan & Kaplan, 1957) and modern Affect Regulation Theory. These models suggest that certain individuals fail to differentiate visceral hunger sensations from physiological correlates of emotional arousal. Consequently, food consumption is instrumentalized as a compensatory coping mechanism to attenuate negative affect, alleviate distress, and achieve acute emotional blunting through rapid neurochemical gratification.

Validity

The psychometric validity of the TFEQ-R18 has been substantiated across diverse clinical, bariatric, and community cohorts internationally, demonstrating robust construct, convergent, predictive, and discriminant validity.

Construct and Structural Validity

Initial construct validation conducted by Karlsson et al. (2000) utilizing the Swedish Obese Subjects (SOS) cohort (N = 1,489) confirmed that the 18 items loaded cleanly onto three distinct factors, successfully eliminating the psychometric clutter and cross-loadings observed in the parent 51-item tool. Subsequent cross-validation in a Swedish general population sample (N = 1,105) corroborated that the three-factor structure operated invariantly across individuals with normal weight, overweight, and obesity.

Convergent and Discriminant Validity

The convergent validity of the TFEQ-R18 is extensively documented through correlations with related psychological and anthropometric constructs. Research by de Lauzon et al. (2004) in a large general population cohort (the French SU.VI.MAX study, N = 5,388) demonstrated that Uncontrolled Eating and Emotional Eating correlate positively with higher body mass index (BMI), elevated daily energy intake, and higher consumption of energy-dense, sweet, and fatty foods. Conversely, Cognitive Restraint correlates with greater dietary fiber intake, lower proportional intake of refined carbohydrates, and adherence to structured dietary programs.

Discriminant validity is supported by low to moderate intercorrelations between subscales. Typically, Cognitive Restraint demonstrates weak or non-significant correlations with Emotional Eating (r ranging from -0.05 to 0.15), confirming that deliberate restriction is structurally distinct from affect-driven eating. Although Uncontrolled Eating and Emotional Eating are moderately correlated (r ranging from 0.50 to 0.65), reflecting shared appetitive vulnerabilities, confirmatory factor analyses consistently demonstrate that collapsing them into a single factor causes substantial deterioration in model fit, justifying their separation.

Predictive and Clinical Validity

The TFEQ-R18 displays exceptional predictive validity in clinical weight management and metabolic bariatric surgery. Prospective studies evaluate longitudinal shifts following Roux-en-Y gastric bypass and sleeve gastrectomy; patients exhibit substantial, rapid reductions in Uncontrolled Eating and Emotional Eating scores alongside modest increases in Cognitive Restraint within the first 12 to 24 months post-surgery. Crucially, post-surgical increases in Uncontrolled Eating or Emotional Eating scores strongly predict weight regain, subclinical loss-of-control eating, and diminished health-related quality of life.

Reliability

The reliability of the TFEQ-R18 has been established through repeated evaluations of internal consistency and temporal stability across diverse populations and cultural adaptations:

Internal Consistency

In the foundational validation study by Karlsson et al. (2000), internal consistency coefficients (Cronbach’s alpha) were robust across both clinical and non-clinical cohorts:

• Cognitive Restraint (CR, 6 items): α = 0.75 in the obese SOS cohort; α = 0.78 in the Swedish general population cohort.
• Uncontrolled Eating (UE, 9 items): α = 0.89 in the obese SOS cohort; α = 0.87 in the general population cohort.
• Emotional Eating (EE, 3 items): α = 0.87 in the obese SOS cohort; α = 0.85 in the general population cohort.

Subsequent international validations have reaffirmed these findings. In the French validation by de Lauzon et al. (2004), Cronbach’s alpha values were 0.76 for CR, 0.84 for UE, and 0.86 for EE. In a validation of the Finnish TFEQ-R18 by Anglé et al. (2009), coefficients were 0.76 (CR), 0.87 (UE), and 0.88 (EE). Across adult populations, composite reliability indices systematically exceed the 0.70 benchmark required for research purposes, with UE and EE consistently meeting the 0.85 standard recommended for individual clinical profiling.

Test-Retest Reliability

Temporal stability evaluated over intervals ranging from 2 weeks to 4 weeks demonstrates high test-retest reliability. Intraclass correlation coefficients (ICC) and Pearson correlation coefficients consistently range between 0.78 and 0.88 for Cognitive Restraint, 0.77 and 0.85 for Uncontrolled Eating, and 0.80 and 0.89 for Emotional Eating. These statistics confirm that the TFEQ-R18 captures stable behavioral traits while retaining adequate sensitivity to detect longitudinal treatment effects following behavioral, pharmacological, or surgical interventions.

Factor Analysis

The factorial validity of the TFEQ-R18 is one of its central psychometric strengths, resolving the profound dimensionality disputes that surrounded its 51-item predecessor. The empirical development and subsequent replication of its factor structure have been rigorously demonstrated through both Exploratory Factor Analysis (EFA) and Confirmatory Factor Analysis (CFA).

Exploratory Factor Analysis (EFA)

Karlsson and colleagues (2000) initiated item reduction by analyzing the responses of 1,489 obese patients in the Swedish Obese Subjects study. Principal components analysis followed by orthogonal (Varimax) and oblique rotations demonstrated that the original 51 items exhibited substantial multidimensionality and cross-loadings. By eliminating items with poor factor loadings (< 0.40), significant secondary cross-loadings (> 0.35 on non-target factors), or severe semantic overlap, the authors derived a parsimonious 18-item structure accounting for 48.7% of the total variance across three cleanly demarcated factors. All 6 items of Cognitive Restraint exhibited primary loadings between 0.52 and 0.76; the 9 Uncontrolled Eating items loaded between 0.48 and 0.77; and the 3 Emotional Eating items exhibited robust factor loadings ranging from 0.74 to 0.86.

Confirmatory Factor Analysis (CFA)

Numerous CFA studies worldwide have confirmed the superiority of the correlated three-factor model over unidimensional or two-factor alternatives. Fit indices across diverse international datasets indicate good to excellent model fit:

• Comparative Fit Index (CFI): Routinely observed between 0.93 and 0.97 (exceeding the standard 0.90 threshold for acceptable fit and frequently approaching the 0.95 benchmark for superior fit).
• Tucker-Lewis Index (TLI): Typically ranges from 0.92 to 0.96.
• Root Mean Square Error of Approximation (RMSEA): Generally spans 0.048 to 0.065, well below the 0.08 cutoff indicating reasonable error of approximation.
• Standardized Root Mean Square Residual (SRMR): Consistently documented below 0.055.

Multigroup confirmatory factor analyses (MGCFA) have further confirmed full metric and scalar measurement invariance across sex (males vs. females) and weight classifications (normal-weight, overweight, and class I–III obesity), verifying that group comparisons reflect genuine psychological variations rather than measurement artifacts.

Instrument / Measurement Tool

• Test Type: Self-administered psychological self-report questionnaire / Patient-Reported Outcome Measure (PROM).
• Target Population: Adults (aged 18 and older); validated in normal-weight, overweight, obese, and post-bariatric surgery populations. (Adolescent versions, such as TFEQ-R18V2 / TFEQ-R17, have also been adapted).
• Administration Format: Paper-and-pencil, computer-based, or mobile digital survey administration.
• Item Count: 18 items.
• Subscale Composition:
  • Cognitive Restraint (CR): 6 items (Items 2, 11, 12, 15, 16, 18).
  • Uncontrolled Eating (UE): 9 items (Items 1, 4, 5, 7, 8, 9, 13, 14, 17).
  • Emotional Eating (EE): 3 items (Items 3, 6, 10).
• Response Scale:
  • Items 1–17: 4-point Likert-type response scale: 1 = Definitely false, 2 = Mostly false, 3 = Mostly true, 4 = Definitely true.
  • Item 18: 8-point numerical rating scale where 1 = no restraint in eating and 8 = total restraint. Responses are recoded into a 4-point scale: responses 1–2 = 1, 3–4 = 2, 5–6 = 3, 7–8 = 4.
• Scoring and Transformation Rules:
  • Raw subscale scores are obtained by calculating the sum of the items assigned to each subscale: Cognitive Restraint (range: 6–24), Uncontrolled Eating (range: 9–36), and Emotional Eating (range: 3–12). Mean subscale scores (range: 1–4) are also frequently reported.
  • Raw scores are routinely transformed to a standardized 0–100 scale using the linear formula:
    Standardized Score = [ (Raw Score - Lowest Possible Raw Score) / (Possible Raw Score Range) ] × 100
  • Higher standardized scores denote higher levels of Cognitive Restraint, Uncontrolled Eating, or Emotional Eating.

Permissions & Fee and Test Year

The Three-Factor Eating Questionnaire – Revised 18 (TFEQ-R18) was published in the year 2000 by Jan Karlsson, Lars-Olof Persson, Lars Sjöström, and Marianne Sullivan in the peer-reviewed journal International Journal of Obesity. The instrument was derived as a public academic contribution within the Swedish Obese Subjects (SOS) study framework. The original publication is copyrighted by Nature Publishing Group / Macmillan Publishers Ltd.

In academic, clinical, and non-commercial research contexts, the TFEQ-R18 is widely accessible and utilized without royalty fees, provided that standard academic citation and ethical acknowledgment are accorded to the original authors and publisher. Commercial entities, pharmaceutical clinical trials, or software developers integrating the scale into proprietary platforms should contact the primary authors or the copyright holder to verify licensing permissions.

References

• Anglé, S., Engblom, J., Eriksson, T., Kautiainen, S., Saha, M. T., Lindfors, P., Lehtinen, R., & Rimpelä, A. (2009). Three-Factor Eating Questionnaire-R18 as a measure of cognitive restraint, uncontrolled eating and emotional eating in a sample of young Finnish adults. International Journal of Behavioral Nutrition and Physical Activity, 6(1), Article 41. https://doi.org/10.1186/1479-5868-6-41
• Bruch, H. (1961). Transformation of oral impulses in eating disorders: A conceptual approach. Psychiatric Quarterly, 35(3), 458–481. https://doi.org/10.1007/BF01566779
• de Lauzon, B., Romon, M., Deschamps, V., Lafay, L., Borys, J. M., Eschwège, E., & Charles, M. A. (2004). The Three-Factor Eating Questionnaire-R18 is able to distinguish among different eating patterns in a general population. The Journal of Nutrition, 134(9), 2372–2380. https://doi.org/10.1093/jn/134.9.2372
• Herman, C. P., & Polivy, J. (1975). Anxiety, restraint, and eating. Journal of Abnormal Psychology, 84(6), 666–672. https://doi.org/10.1037/0021-843X.84.6.666
• Kaplan, H. I., & Kaplan, H. S. (1957). The psychosomatic concept of obesity. The Journal of Nervous and Mental Disease, 125(2), 181–201. https://doi.org/10.1097/00005053-195704000-00004
• Karlsson, J., Persson, L. O., Sjöström, L., & Sullivan, M. (2000). Psychometric properties and factor structure of the Three-Factor Eating Questionnaire (TFEQ) in obese men and women. Results from the Swedish Obese Subjects (SOS) study. International Journal of Obesity and Related Metabolic Disorders, 24(12), 1715–1725. https://doi.org/10.1038/sj.ijo.0801442
• Schachter, S. (1968). Obesity and eating. Science, 161(3843), 751–756. https://doi.org/10.1126/science.161.3843.751
• Stunkard, A. J., & Messick, S. (1985). The three-factor eating questionnaire to measure dietary restraint, disinhibition and hunger. Journal of Psychosomatic Research, 29(1), 71–83. https://doi.org/10.1016/0022-3999(85)90010-8

Items of the Scale