Prediction (regression) equations are widely used, but their reliability as predictive tools is questionable as they provide contradicting results. The key point is that values calculated by regression equations are not precisely defined numbers but lie within a range of possible values in the standard deviation interval, none of which can be considered as the most probable. Ignoring this point leads to illicit/improper calculations, generating wrong results, which may have adverse consequences for human health. To demonstrate this, we applied the equations of Harris and Benedict in a reverse method, i.e. calculating (predicting) the daily energy expenditure in the same subjects used to obtain the equations and comparing values with the original measured data. We used the Bland-Altman and frequency distribution analyses. We found large differences in both individual data and population characteristics, showing that prediction equations are not predictive tools.