The paper introduces RECALL (Relative Conditional Log-Likelihood), a novel membership inference attack (MIA) designed to detect pretraining data in large language models (LLMs). RECALL leverages the conditional language modeling capabilities of LLMs by examining the relative change in conditional log-likelihoods when prefixing target data points with non-member context. The key finding is that conditioning member data on non-member prefixes induces a larger decrease in log-likelihood compared to non-member data. Empirical results show that RECALL achieves state-of-the-art performance on the WikiMIA dataset, even with random and synthetic prefixes, and can be further improved using an ensemble approach. The paper also provides an in-depth analysis of LLMs' behavior with different membership contexts, offering insights into how LLMs leverage membership information for effective inference at both the sequence and token levels.The paper introduces RECALL (Relative Conditional Log-Likelihood), a novel membership inference attack (MIA) designed to detect pretraining data in large language models (LLMs). RECALL leverages the conditional language modeling capabilities of LLMs by examining the relative change in conditional log-likelihoods when prefixing target data points with non-member context. The key finding is that conditioning member data on non-member prefixes induces a larger decrease in log-likelihood compared to non-member data. Empirical results show that RECALL achieves state-of-the-art performance on the WikiMIA dataset, even with random and synthetic prefixes, and can be further improved using an ensemble approach. The paper also provides an in-depth analysis of LLMs' behavior with different membership contexts, offering insights into how LLMs leverage membership information for effective inference at both the sequence and token levels.