Aims: Vectorcardiography (VCG) is generated by projecting signals from several leads onto three main orthogonal axes. There is evidence that by doing this projection, some relevant diagnostic information may be lost. We investigated a new way to reduce this information loss. We generated a 12-dimensional VCG (VCG12D) from the standard 12-lead ECG system and compared its performance in diagnosing myocardial infarction (MI) with standard Frank VCG (VCGFrank) and with a three-dimensional projection of the 12-lead ECG obtained with principal component analysis (VCGPCA). Methods: All 148 MI patients and 52 healthy controls from the PTB Diagnostic ECG Database were selected. Standard 12-lead ECG and VCGFrank were available for all subjects. We generated VCG12D using all the leads of the standard 12-lead ECG and associating them with coordinates in a 12-dimensional space. A set of standard parameters in the VCG literature (loop-area, loop-perimeter, etc.) were computed for VCG12D, VCGFrank, and VCGPCA. Then, single feature logistic regression was used to assess performance of each individual parameter in diagnosing MI, for VCG12D, VCGPCA and VCGFrank, respectively. Additionally, a multivariate lasso regression model was generated for VCG12D, VCGPCA and VCGFrank, respectively, by using all parameters as initial input. Results: When diagnosing the MI condition with single feature logistic regression, the best single feature performances for VCGFrank and VCG12D were comparable, having AUC of 0.94 and 0.95 respectively. Performance for VCGPCA was poorer, with AUC of 0.81 for its best feature. When using lasso, AUC was 0.93 for VCGFrank, 0.97 for VCG12D and 0.90 for VCGPCA. Conclusion: VCG12D performs better than VCGPCA and not significantly better than VCGFrank. This partially confirms that projecting information into 3D may cause a loss of diagnostically relevant information. Nonetheless, differences between VCG12D and VCGFrank are not significant and more research is needed to investigate potential benefits of a multidimensional VCG.