Although Heart Rate Variability (HRV) was shown to be associated with mortality of post myocardial infarction (MI) patients, the clinical use of conventional HRV parameters is limited. We proposed new method based on the analysis of autocorrelation function (ACF) of detrended RR interval recordings. One thousand four hundred fifty five consecutive post MI patients were enrolled into the prospective blinded study at German Heart Center in Munich. The RR intervals were extracted from 24 h Holter recordings using integrated software. The linear trend was estimated and subtracted from RR recordings. The ACF(m) was calculated for original and detrended RR-intervals. The slope (AS) of regression line fitted for ACF (m; m=1,10) was dichotomized at AS=-0.0025. AS >-0.0025 was considered as pathological. The significant association of pathological AS was revealed in univariate analysis (Log Rank =39.19, p