The amount and distribution of adipose tissue (AT) in humans plays an important role in the pathogenesis of metabolic diseases. It is well known that especially visceral adipose tissue (VAT) is increased in insulin resistant subjects. Recently it has been shown that it is possible to identify a form of benign adiposity in obese subjects with body mass index (BMI) >30kg/m² [1] with lower amounts of VAT and intrahepatic lipids (IHL) for the quartile with high insulin sensitivity index (ISI). Aim of this analysis was to identify metabolically unhealthy non-obese subjects – thin outside, fat inside (TOFI) – and to characterize them regarding distribution of whole-body adipose tissue.Four hundred ninety-eight non-obese subjects (f/m 312/186) underwent an initial MR examination prior to participating in a lifestyle intervention program. All subjects were at increased risk for metabolic diseases due to being overweight (BMI>27 but lower 30 kg/m²), being a first degree relative of a patient with type 2 diabetes, having impaired glucose tolerance and/or a gestational diabetes. MR examinations were performed on a 1.5 T whole-body imager (Magnetom Sonata, Siemens Healthcare, Germany) in the early morning after overnight fasting including whole-body T1-weighted MRI for segmentation of adipose- and lean tissue compartments [2] as well as 1H-MRS of the liver for quantification of IHL by a single voxel STEAM technique [3]. AT of the lower legs (AT_LE), of the trunk excluding VAT (NVAT_T), VAT and of the upper extremities (AT_UE) was quantified by automatic segmentation [4] and is expressed as percentage of total body adipose tissue (TAT). Lean tissue compartments were assessed from lower extremities (LT_LE), trunk (LT_T) and upper extremities including head (LT_UE). Anthropometrics and insulin sensitivity – determined by an oral glucose tolerance test – were assessed immediately after the MR exam. Subjects were divided in an insulin sensitive (IS) and an insulin resistant (IR) group by the median of ISI for further classifications.Table 1 lists the anthropometric, metabolic and MR-derived parameters for the entire cohort for males and females seperately as well as the IS and IR groups. The IR groups are slightly older compared to the IS counterparts and have a significantly higher BMI which is, however, far away from obesity. Gender related differences in AT distribution are obvious for all compartments but there are no differences in LT distribution reflecting muscle mass in the extremities and muscle+organ tissue in the trunk. Regarding AT distribution of IS and IR subjects there are significant differences with higher %VAT and %NVAT_T but lower %AT_LE and %AT_UE (significant for females, almost identical for males) in the IR groups. IHL are almost doubled for IR subjects. Figure 1 shows exemplary T1-weighted images of the umbilical level and thighs of an IS (a) and an IR female (b) as well as for a male with corresponding spectra of the liver (c: IS; d: IR).The special role of different AT compartments and their meaning in the progress of metabolic diseases is undoubted. However, besides obesity which is a primary risk factor for metabolic syndrome and/or type 2 diabetes, also non-obese subjects can develop a prediabetic state, i.e. (yet) reversible insulin-resistance. The results of this cross-sectional analysis show marked and significant differences in AT distribution with a more pronounced accumulation in extremities for IS non-obese subjects. This reflects the apple- and pear-shape which does not only differs for males and females but also for IS and IR subjects in the same gender. These data – not taking into consideration the total amount of AT compartments – allow a detailed MR-based phenotyping of subjects at increased risk for metabolic diseases and might enable an identification of subjects which are non-responders to lifestyle intervention including dietary changes and increased exercise [5]. It has to be mentioned, that our results do not reflect the general population – however, the differences might even be more pronounced in subjects not fulfilling the mentioned inclusion criteria.Thin outside, fat inside – generally it is not possible to identify people at risk for metabolic diseases just by visual inspection or simple anthropometrics as waist circumference or bioimpedance analysis. MRI and MRS offer a unique possibility for differentiation of these phenotypes. Our results regarding adipose tissue distribution highlight significant disparities with a shift towards AT accumulation in the extremities for metabolically healthy subjects and might help to define cut-off values for single compartments and transfer to epidemiological studies in large cohorts.