The analysis of the T-cell receptor (TCR) is becoming an increasingly important tool in the study of the tumor microenvironment. Global features of the repertoire, including clonality and diversity, can give important insights into the immunological processes which drive cancer progression or regression, especially in the context of immune checkpoint inhibitors. Specific TCR sequences are increasingly being used as part of the computational or experimental antigen discovery pipeline. However, the clinical samples available from tumors frequently contain poor-quality nucleic acids, making the amplification of TCR libraries challenging. We describe a robust and inexpensive protocol which can be used in conjunction with our computational TCR annotation analysis pipeline, Decombinator, to generate TCR repertoires from a variety of RNA sources and concentrations, including poor-quality low-concentration RNA from formalin-fixed paraffin-embedded tissue. We call the method (FFPE-suitable Unique Molecular idEntifier-based TCRseq (FUME-TCRseq)). The method uses a multiplex PCR to amplify the TCRbeta CDR3 region, together with sufficient V and J gene sequence to allow accurate annotation. The method incorporates a 12-base pair unique molecular identifier into each cDNA molecule, allowing downstream correction for sequencing errors, as well as PCR bias. The method unlocks a vast collection of archived tumor samples for TCR repertoire analysis, potentially leading to a better understanding of the immunology of the tumor/host response, and the mode of action of immunotherapeutic intervention.