The Minnesota Multiphasic Personality Inventory-2 (MMPI-2) is a popular measure of psychosocial functioning and psychopathology in the assessment of individuals in a variety of settings. However, the method of construction employed with the MMPI more than 60 years ago with psychiatric patients challenges the applicability of the scales for determining the psychosocial functioning of individuals from different settings. The restandardisation conducted in 1987 made no effort to eradicate the item overlap that was a result of the criterion keying method with contrasted groups. Although restandardized and updated with more contemporary language and content, the original psychiatric constructs were retained in order to maintain continuity with its predecessor. The aims of this investigation were to develop a new structure for the MMPI-2 constructed at the item-level, empirically derived and which specifically represents the dimensions that are relevant and appropriate in evaluating the psychosocial functioning of personal injury claimants. This task included comparisons with a comparable scale-level analysis and developing optimal scoring strategies where items in components and facets are allocated weightings based upon their strength of association. Study 1 was conducted using a sample of 2989 personal injury claimants assessed in Australia and the United States of America. The final sample of 3230, included 241 normal individuals, was utilized to develop a scale-level structure from 79 standard MMPI-2 scales and subscales. A nine-component solution consisting of General Maladjustment /Emotional Distress, Asocial Beliefs, Social Vulnerability, Somatic Complaints, Psychological Disturbance, Impulsive Expression, Antisocial Practices, Stereotypic Fears and Family Difficulties was derived using principal component analysis. However, intercorrelation between components in the structure signaled the need to develop a structure that would eradicate problems that were perpetuated by item overlap. The second study was conducted with a set of best practice procedures with the same clinical sample of 2989 personal injury claimants as Study 1. Forty-one components were derived through principal component analysis. Through the application of a set of criteria, a 35-component solution was retained. The pattern coefficients from the allocation of items to components determined the weightings to be applied to each item. Further analysis of the 35 components derived a substructure of 37 facets. The 35 components included only 442 of the 567 items, with the reliability coefficients of the first 25 components that ranged between .5 and .97, and the remaining 10 components that ranged from .29 to .49. The latter unreliable components were not included in the final Structural Summary, leaving 25 components (400 items) and their 33 facets for interpretation. Hence, in demonstrating the utility of the newly-derived structure, only 25 components and their 33 facets were interpreted. The 25 components were grouped conceptually into six domains. In the emotional domain were Psychological Distress (PsyDist), Anger, Fears, Psychotic Symptoms (PsyS), Paranoia (Par), Irritability (Irrit), Elation (Elat), Fear of the Dark (FD), and Financial Worry (FinWo). Somatic Complaints (SomC), Sexual Concerns (SexCon), and Gastrointestinal Problems (GasP) made up the measures in the physiological domain. In the behavioural domain were Cognitive Difficulties (CogDiff), Stimulus-Seeking (StimuS), Discipline (Dis), and Delinquency (Del) whilst the interpersonal domain was formed by Social Withdrawal (SoW), Negative Interpersonal Attitude (NIA), Timidity (Tim), Lie, Dissatisfaction with Self (DWS) and Family Relationship Difficulties (FReD). Alcoholism (Alco) was the only measure in the substance abuse domain, and the gender domain was comprised of Masculinity (Mas) and Femininity (Fem). The third study established preliminary normative means and standard deviations using a small opportunistic Australian university student sample (N = 219). No substantial gender differences were found but gender norms were maintained to facilitate comparisons with the traditional MMPI-2 approach. Comparisons of frequency of 'true' item response between the Australian university student sample and the U.S. restandardisation sample found relatively little differences and permitted evaluation of between sample differences on components and facets. The utility of the structure was demonstrated with the illustration of two clinical case examples, and a comparison was made with the standard MMPI-2 scales and subscales. The Structural Summary for the MMPI-2 demonstrated discriminative measures of psychosocial functioning that were a result of no item overlap, and the ability to attend to the different levels of intensity of self-report items because of differential weightings.