The diagnosis of Lyme disease is difficult because tests that reflect active disease or have reasonable sensitivity and specificity are lacking or not timely. Molecular methods are controversial because of differences in assays, gene targets, and limited clinical validation. This review summarizes published assays for Lyme disease diagnosis using skin, plasma, synovial fluid, cerebrospinal fluid (CSF), and urine. Meta-analyses show the strengths and weaknesses of these methods. Overall, assays for skin and synovial fluid (68% and 73%, respectively) have high sensitivity and uniformity. The low test sensitivity of CSF (18%) and plasma (29%), variable sensitivities among CSF and urine assays, and persistence of Borrelia burgdorferi DNA in urine and synovial fluid even with therapy and convalescence make these unsuitable for primary diagnosis. Molecular assays for Lyme disease are best used with other diagnostic methods and only in situations in which the clinical probability of Lyme disease is high.