The prediction of batteries final failure in standalone photovoltaic systems is a quite important issue. The implementation of a simple and efficient method for such systems is mandatory for any application. The scale parameter of the discharge voltage distribution Weibull fit over time and its corresponding Root Mean Square Error (with respect to the initial distribution) evolution are proposed, checked and compared as useful tools for the early prediction of batteries breakdowns. The experimental set-up is a macro-grid of 1600 identical stand alone PV systems. (Pierre Auger Observatory). But not only is the prediction of the batteries lifetime important. The indication of the possible cause of dead is an interesting point. The sensitivity for different basic parameters intrinsically related to stand-alone photovoltaic systems (weather during installation, time of storage, kind of anomaly‥) was also studied for both algorithms.