Nostalgia is a self‐conscious social emotion that arises from reminiscence of past memories. Collective nostalgia on social media such as Twitter has been seen as a method to comfort individuals in the status of isolation, fear, and a loss of freedom. In recent years, many studies have focused on offering analysis of nostalgic conversations to understand their impact on various domains including marketing and mental health, but little attention has been given to how to detect such conversations in the first place. This paper offers a novel large‐scale nostalgic tweets dataset. We describe our extensive methodology to create and validate this dataset using natural language processing models and Large Language Models to detect nostalgic conversations on Twitter. We demonstrate the properties of this dataset alongside analysis revealing insight into context and patterns of what/how people reminiscence about. We finish the paper by describing other research studies that our dataset enables.