|Kurzfassung :||The focus of the present work firstly, was on the theoretical and methodological substantiation of social intelligence, and secondly on the development of a test battery of social intelligence (i.e., the Social Intelligence Test Magdeburg, SIM). Social intelligence, in the present study, is based on the model of Weis and Süß (2005; see also Weis, Seidel, & Süß, 2006) and only contains cognitive ability requirements. The model originally represented a structural model distinguishing between social understanding, social memory, social perception, and social creativity as the cognitive ability domains. The model is modified in the present work by adding a hierarchical assumption in terms of a higher-order social intelligence factor. In extension to this differentiation of operative requirements, further taxonomic considerations are applied which identify additional relevant classificatory principles in definitions of social intelligence: (a) contents or cues, (b) the queried modalities, (c) the settings, and (d) the targets. These taxonomic considerations serve as classificatory principles of existing measurement approaches and as foundations for the subsequent test development. With respect to the empirical background, existing measurement approaches and the surrounding methodological problems are described and discussed in light of the methodological shortcomings and the resulting validity evidence. The conclusions identify problems surrounding the use of artificial and decontextualized item material, the use of only written language contents, the adequate scoring procedure, and a mismatch between the purported measurement construct and the actual task requirements. As a consequence, many existing measurement approaches lack evidence for the convergent and divergent construct validity. Consequently, the test development in the present work was based on the model of Weis and Süß (2005) and the associated taxonomic considerations. The test design cross-classified three operative ability domains (i.e., social understanding, social memory, and social perception) and four material related content domains (i.e., written and spoken language, pictures, and videos). Additionally, all tasks systematically varied the type of setting (i.e., private vs. public) and the number of persons involved in the situations (i.e., one person, a dyad, and small groups). All tasks relied on genuine task material sampled in natural settings involving real persons. Amongst others, the pivotal research questions concerned the investigation of the psychometric properties of the social intelligence tasks and the examination of the construct validity of social intelligence as assessed by the SIM. The hypotheses of the construct validity established testable models of (a) the internal structure of social intelligence as postulated in the design of the test battery and (b) the relationship between the broad ability factors of social and academic intelligence (divergent construct validity). Additionally, it was expected that social intelligence would show divergent construct validity with personality traits. The present work was based on two main studies. One hundred twenty six German university students participated in Study 1. The mean age was 21.35 (sd = 3.06), and 53.5 % were females. In Study 2, an unselected sample of adults was applied. Participants were between 23 and 40 years old (mage = 28.69; sd = 5.57) and 58.8 % of the subjects were female. The sample consisted of heterogeneous subjects in terms of age, education, and occupation. Both studies applied the test battery of social intelligence and the Berlin Intelligence Structure Test (BIS-Test; Jäger, Süß, & Beauducel, 1997) as a measure of academic intelligence. Additionally, several trait inventories of personality were applied. Only some of the main results will be pointed out in this abstract: Referring to the psychometric properties, most of the newly developed tasks showed sufficient reliability coefficients. After some necessary steps of data cleaning (i.e., dealing with missing values, correction of outliers, and trimming of the reaction time scores), the final scales were found to be normally distributed. With respect to the internal structure of social intelligence, confirmatory factor analysis supported a two-factor structural model with two correlated factors of social understanding and memory. Regarding the construct validity, confirmatory factor analysis supported discriminable social and academic intelligence factors with low correlations between social understanding and the BIS-Reasoning factor. The social memory factor and BIS-Memory were substantially correlated. However, the social intelligence structural model proved structure independency from academic intelligence: when BIS-Test variance was partialled out of the single tasks, confirmatory factor analysis replicated the structural model of social intelligence based on the residuals. Furthermore, correlational analysis supported the divergent construct validity of the social intelligence tasks with personality traits. Besides the discussion of the aforementioned research results and their consequences for the social intelligence construct, the test approach and the underlying taxonomic considerations are discussed. Several decisions during the course of test development and the resultant tasks are subject to critical discussion such as the application of genuine task material, the target scoring procedure, the process of item construction and selection, and the cognitive processes underlying the social understanding tasks. Finally, some considerations about test modifications and extensions are addressed. Complementary interesting research questions that add to the current results are also discussed.