Deciphering the molecular pathogenesis of virally induced cancers is challenging due, in part, to the heterogeneity of both viral gene expression and host gene expression. Epstein-Barr virus (EBV) is a ubiquitous herpesvirus prevalent in B-cell lymphomas of immune-suppressed individuals. EBV infection of primary human B cells leads to their immortalization into lymphoblastoid cell lines (LCLs), serving as a model of these lymphomas. In previous studies, reports from our laboratory have described a temporal model for immortalization with an initial phase characterized by expression of Epstein-Barr nuclear antigens (EBNAs), high levels of c-Myc activity, and hyperproliferation in the absence of the latent membrane proteins (LMPs), called latency IIb. This is followed by the long-term outgrowth of LCLs expressing the EBNAs along with the LMPs, particularly NFκB-activating LMP1, defining latency III. However, LCLs express a broad distribution of LMP1 such that a subset of these cells express LMP1 at levels similar to those seen in latency IIb, making it difficult to distinguish these two latency states. In this study, we performed mRNA sequencing (mRNA-Seq) on early EBV-infected latency IIb cells and latency III LCLs sorted by NFκB activity. We found that latency IIb transcriptomes clustered independently from latency III independently of NFκB. We identified and validated mRNAs defining these latency states. Indeed, we were able to distinguish latency IIb cells from LCLs expressing low levels of LMP1 using multiplex RNA-fluorescencein situhybridization (RNA-FISH) targeting EBVEBNA2orLMP1and humanCCR7orMGST1. This report defines latency IIb as a bona fide latency state independent from latency III and identifies biomarkers for understanding EBV-associated tumor heterogeneity.IMPORTANCEEBV is a ubiquitous pathogen, with >95% of adults harboring a life-long latent infection in memory B cells. In immunocompromised individuals, latent EBV infection can result in lymphoma. The established expression profile of these lymphomas is latency III, which includes expression of all latency genes. However, single-cell analysis of EBV latent gene expression in these lymphomas suggests heterogeneity where most cells express the transcription factor, EBNA2, and only a fraction of the cells express membrane protein LMP1. Our work describes an early phase after infection where the EBNAs are expressed without LMP1, called latency IIb. However, LMP1 levels within latency III vary widely, making these states hard to discriminate. This may have important implications for therapeutic responses. It is crucial to distinguish these states to understand the molecular pathogenesis of these lymphomas. Ultimately, better tools to understand the heterogeneity of these cancers will support more-efficacious therapies in the future.