In this first part of our reply to the two papers by Mulargia and Gasperini [1996a,b> we summarize the main deficiencies due to which their present calculations as well as their earlier calculations in 1992 are incorrect beyond any doubt. Our answers, point by point, to other questions raised by Mulargia and Gasperini [1996a,b> are given in the following paper hereafter cited as Reply II. Concerning the calculation of Mulargia and Gasperini [1992>: a) It suffers from a very serious mistake indicated by Rhoades and Evison [1996>, as a result of which Mulargia and Gasperini treated in their calculations probability values as high as 11 (which violate the definition of probability). What they did is akin to approximating the probability of getting heads in at least one of n coin-tosses by the expected number of heads in n tosses; however, the probability only equals the expected value when n=1. b) It did not take into account that each VAN prediction had a certain spatial extent and c) when applied to a fictitious Ideally Perfect Earthquake Prediction Method [IPEPM, which by definition achieves, in a whole geographic area, the prediction of all earthquakes -above a certain magnitude (e.g., MS≥5.3)-and does not issue any false alarms> it leads to a paradox, i.e., to values of the significance level indicating that these ideal predictions can be ascribed to chance. Mulargia and Gasperini [1996a,b> do not provide any support to their earlier work in 1992 and accept their aforementioned mistakes. However, they now claim that they find a significant association between VAN predictions and earthquakes (EQs) in ''reverse time.'' We show that (beyond obvious mistakes in Table 1 of Mulargia and Gasperini [1996a>) this claim is simply due to the fact that, although they use a Poisson distribution, they violate the principle that time-dependent events should not be entered in their calculation. (For example, for MEQ≥5.3, when including time-dependent events we find 6 forward associations and 6 backwards; however, when considering only independent events, there are 5 associations in forward time and only 2 in reverse time). Simplified examples demonstrate that Mulargia and Gasperini's [1996a> procedure leads to a second paradox, i.e., it can ''extract'' a backwards time association even for an IPEPM which can successfully predict both main shocks and (related) aftershocks. Furthermore, Mulargia and Gasperini [1996a> make an erroneous assumption which, in addition to the violation of the true meaning of the experimental error, artificially decreases the VAN success rate by a factor of 2 or so. The main conclusion remains: Once Mulargia and Gasperini's statistical procedure leads to erroneous results when applied (with the same numbers) to an ideal case, it should not have been used as a tool for criticizing an experimental method. ¿ American Geophysical Union 1996