Post-marketing drug surveillance is a critical component of drug safety. Drug regulatory agencies such as the U.S. Food and Drug Administration (FDA) rely on voluntary reports from health professionals and consumers contributed to its FDA Adverse Event Reporting System (FAERS) to identify adverse drug events (ADEs). However, it is widely known that FAERS underestimates the prevalence of certain adverse events. Popular patient social media sites such as DailyStrength and PatientsLikeMe provide new information sources from which patient-reported ADEs may be extracted. In this study, we propose an analytical framework for extracting patient-reported adverse drug events from online patient forums. We develop a novel approach - the AZDrugMiner system - based on statistical learning to extract ad-verse drug events in patient discussions and identify reports from patient experiences. We evaluate our system using a set of manually annotated forum posts which show promising performance. We also examine correlations and differences between patient ADE reports extracted by our system and reports from FAERS. We conclude that patient social media ADE reports can be extracted effectively using our proposed framework. Those patient reports can reflect unique perspectives in treatment and be used to improve patient care and drug safety.