This review discusses the methodology, rationale, design, reporting, and interpretation of disproportionality analyses (DA) derived from spontaneous reporting systems (SRS) in pharmacovigilance. DA is a recognized approach for early signal detection in post-marketing surveillance. While DA cannot replace clinical judgment, it remains essential for detecting rare and unpredictable adverse drug reactions (ADRs) with strong drug-attributable components, especially when conducted by a multidisciplinary team and combined with case-by-case analysis. Recent increases in DA publications have raised concerns about the quality and interpretation of results, with some studies misrepresenting findings to infer causality or risk stratification. The development of international guidelines by the READUS-PV project aims to improve the reproducibility and transparency of DA publications, supporting their use by regulators and clinicians. DA methods include frequentist and Bayesian approaches, such as the Proportional Reporting Ratio (PRR), Reporting Odds Ratio (ROR), and Information Component (IC). These methods use 2x2 contingency tables to compare observed and expected reporting frequencies of drug-event combinations. The Multi-item Gamma-Poisson-Shrinker (MGPS) algorithm is an empirical Bayesian method that adjusts relative reporting ratios. DA findings require careful interpretation, as they cannot estimate risks or confirm causality but can identify AEs with higher-than-expected reporting frequencies. DA findings should be interpreted with caution, considering alternative explanations and biases. Signal validation is crucial to ensure the novelty and strength of evidence for a suspected ADR. DA results should be distinguished from risk of harm, which refers to the probability and severity of harm. SDRs are probabilistic measures of reporting frequencies, while risks of harm are probabilistic estimates of ADR occurrence in a target population. DA findings should be reported as SDRs, not as risks, to ensure transparency and avoid misinterpretation. DA is used in the early stages of signal detection in SRS databases. Signal validation involves verifying the novelty of the suspected ADR and ensuring the statistical findings are robust. DA findings should be combined with clinical and pharmacological assessments to determine the novelty and strength of evidence for a suspected ADR. DA is particularly useful for detecting rare, unexpected, late-onset, or long-lasting ADRs that may not be fully appreciated during pre-marketing phases. DA can also be combined with other data sources, such as electronic health records (EHRs) or in vitro/ex vivo assays, to corroborate biological plausibility and enhance signal detection. DA should not be used as a direct measure of risk, as it cannot account for confounding factors or provide a denominator for drug usage. DA is not suitable for comparing the safety profiles of different drugs, as it is subject to biases inherent in spontaneous reporting data. DA should not be used to describe already well-known ADRs, as this may be redundant or misleading. DA should be used when descriptive analyses would be sufficient, as it may notThis review discusses the methodology, rationale, design, reporting, and interpretation of disproportionality analyses (DA) derived from spontaneous reporting systems (SRS) in pharmacovigilance. DA is a recognized approach for early signal detection in post-marketing surveillance. While DA cannot replace clinical judgment, it remains essential for detecting rare and unpredictable adverse drug reactions (ADRs) with strong drug-attributable components, especially when conducted by a multidisciplinary team and combined with case-by-case analysis. Recent increases in DA publications have raised concerns about the quality and interpretation of results, with some studies misrepresenting findings to infer causality or risk stratification. The development of international guidelines by the READUS-PV project aims to improve the reproducibility and transparency of DA publications, supporting their use by regulators and clinicians. DA methods include frequentist and Bayesian approaches, such as the Proportional Reporting Ratio (PRR), Reporting Odds Ratio (ROR), and Information Component (IC). These methods use 2x2 contingency tables to compare observed and expected reporting frequencies of drug-event combinations. The Multi-item Gamma-Poisson-Shrinker (MGPS) algorithm is an empirical Bayesian method that adjusts relative reporting ratios. DA findings require careful interpretation, as they cannot estimate risks or confirm causality but can identify AEs with higher-than-expected reporting frequencies. DA findings should be interpreted with caution, considering alternative explanations and biases. Signal validation is crucial to ensure the novelty and strength of evidence for a suspected ADR. DA results should be distinguished from risk of harm, which refers to the probability and severity of harm. SDRs are probabilistic measures of reporting frequencies, while risks of harm are probabilistic estimates of ADR occurrence in a target population. DA findings should be reported as SDRs, not as risks, to ensure transparency and avoid misinterpretation. DA is used in the early stages of signal detection in SRS databases. Signal validation involves verifying the novelty of the suspected ADR and ensuring the statistical findings are robust. DA findings should be combined with clinical and pharmacological assessments to determine the novelty and strength of evidence for a suspected ADR. DA is particularly useful for detecting rare, unexpected, late-onset, or long-lasting ADRs that may not be fully appreciated during pre-marketing phases. DA can also be combined with other data sources, such as electronic health records (EHRs) or in vitro/ex vivo assays, to corroborate biological plausibility and enhance signal detection. DA should not be used as a direct measure of risk, as it cannot account for confounding factors or provide a denominator for drug usage. DA is not suitable for comparing the safety profiles of different drugs, as it is subject to biases inherent in spontaneous reporting data. DA should not be used to describe already well-known ADRs, as this may be redundant or misleading. DA should be used when descriptive analyses would be sufficient, as it may not