A new tolerance factor, τ, has been developed to predict the stability of perovskite oxides and halides with high accuracy. This factor is derived using a data analytics approach based on SISSO (Sure Independence Screening and Sparsifying Operator) and is shown to correctly predict the stability of 92% of 576 experimentally characterized ABX₃ compounds. τ outperforms the traditional Goldschmidt tolerance factor, t, in accuracy and generalization, particularly for compounds containing heavier halides. τ is a one-dimensional descriptor that depends on oxidation states and Shannon ionic radii, and it provides a monotonic estimate of the probability that a material is stable as a perovskite. This allows for the identification of 23,314 new double perovskites, ranked by their probability of being stable. τ also demonstrates strong agreement with density functional theory (DFT) calculations for the stability of perovskite compounds, showing linear correlation between τ-derived probabilities and DFT-computed decomposition enthalpies. The new tolerance factor is applicable to both single and double perovskites and has been shown to generalize well to a wide range of compositions. The results highlight the potential of τ as a reliable tool for predicting perovskite stability and guiding the discovery of new materials for various applications, including photovoltaics and electrocatalysis.A new tolerance factor, τ, has been developed to predict the stability of perovskite oxides and halides with high accuracy. This factor is derived using a data analytics approach based on SISSO (Sure Independence Screening and Sparsifying Operator) and is shown to correctly predict the stability of 92% of 576 experimentally characterized ABX₃ compounds. τ outperforms the traditional Goldschmidt tolerance factor, t, in accuracy and generalization, particularly for compounds containing heavier halides. τ is a one-dimensional descriptor that depends on oxidation states and Shannon ionic radii, and it provides a monotonic estimate of the probability that a material is stable as a perovskite. This allows for the identification of 23,314 new double perovskites, ranked by their probability of being stable. τ also demonstrates strong agreement with density functional theory (DFT) calculations for the stability of perovskite compounds, showing linear correlation between τ-derived probabilities and DFT-computed decomposition enthalpies. The new tolerance factor is applicable to both single and double perovskites and has been shown to generalize well to a wide range of compositions. The results highlight the potential of τ as a reliable tool for predicting perovskite stability and guiding the discovery of new materials for various applications, including photovoltaics and electrocatalysis.