Drought periods appear as extreme events whose characterization encompasses several issues. Many issues are related to the multiple natures that a drought may have: meteorological, hydrological, agricultural, or socioeconomical. There are also issues related to the complexity of the phenomena, which may be characterized by many magnitudes, such as duration, severity, or intensity. None of them alone may be used as a general drought characterization criterion. Others arise from the kind of methodologies that are available for their significance evaluation, which focus on different aspects for specific objectives. However, most have a common aspect: the extreme persistent realization of a random hydroclimatic variable. For the goal of general drought analysis, in this paper a new index for drought characterization is presented: the drought frequency index (DFI). The index focuses on this common aspect of the drought origins, with a purely probabilistic treatment. Because droughts are persistent phenomena, the index is based on the stochastic characterization of extreme persistent deviation sequences using a novel probabilistic criterion. In this way, the DFI is related to the mean frequency of recurrence of extreme persistent events. Therefore the mean frequency of recurrence is adopted as the scale for drought significance evaluation. The index performance is analyzed and compared with respect to the different issues that result from applying other methodologies: magnitude selection, univariate versus multivariate, threshold selection (related to the run theory), and timescale issues (related with the standard precipitation index (SPI) application). Furthermore, to apply runs theory for any magnitude number, an original generalization of drought multivariate recurrence models is presented. Finally, the spatial comparability of the indexes is analyzed. Results reveal the ability of the DFI to reduce the sensitivity to practical issues. The DFI provides a consistent index for spatial comparisons and for application to general drought characterization goals.