Liquid biopsies detect disease noninvasively by profiling cell-free nucleic acids that are secreted into the circulation. However, existing methods exhibit low sensitivity for detecting early stages of diseases such as cancer. Here we show that long-read nanopore sequencing of full-length cell-free RNA in plasma from healthy individuals, precancerous Barretts esophagus patients with high-grade dysplasia, or patients with esophageal adenocarcinoma reveals a diverse cell-free RNA transcriptome that can be leveraged for detecting and treating disease. We discovered 270,679 novel, intergenic cell-free RNAs, which we used to build a custom transcriptome reference for quantification, feature selection, and machine learning to accurately classify both precancer and cancer. Moreover, we found potential therapeutic targets, including metabolic, signaling, and immune checkpoint pathways, that are highly upregulated in both precancer and cancer patients. Our findings highlight the utility of our RNA liquid biopsy platform technology for discovering and targeting early stages of disease with molecular precision.