The preanalytical handling of plasma, how it is drawn, processed, and stored, influences its composition. Samples in biobanks often lack this information and, consequently, important information about their quality. Especially metabolite concentrations are affected by preanalytical handling, making conclusions from metabolomics studies particularly sensitive to misinterpretations. The perturbed metabolite profile, however, also offers an attractive choice for assessing the preanalytical history from the measured data. Here we show that it is possible using Orthogonal Projections to Latent Structures Discriminative Analysis to divide plasma NMR data into a multivariate "original sample space" suitable for further less biased metabolomics analysis and an orthogonal "preanalytical handling space" describing the changes occurring from preanalytical mishandling. Apart from confirming established preanalytical effects on metabolite levels, e.g., the consequent changes in glucose, lactate, ornithine, and pyruvate, the sample preparation protocol involved methanol precipitation which allowed the observation of reversible changes in short-chain fatty acid concentrations as a function of temperature.