Definite diagnosis of Lyme neuroborreliosis (LNB) requires investigation of serum and cerebrospinal fluid (CSF). Thus, lumbar puncture is necessary, and requires administration of sedating drugs in children. This study aimed to investigate if a pattern of different inflammatory biomarkers in serum could contribute to the selection of children for lumbar puncture in suspected LNB. Patients were included from a cohort of children who was previously investigated for LNB including serum and CSF sampling during the years 2010-2014. The multiplex proximity extension assay (PEA) inflammation panel Target 96 (Olink Bioscience, Uppsala, Sweden) was used to examine 92 biomarkers in serum. Based on the presence of CSF pleocytosis and Borrelia-specific antibodies, patients were divided into a definite LNB group (n=61) and a non-LNB control group (n=58). Following PEA and statistical analysis with multivariate logistic regression, five biomarkers remained significant (p < 0.001), which were included in a calculation of protein index. The index biomarkers were CST5, IL-15RA, CXCL10, DNER and CX3CL1. A receiver operating characteristic curve was constructed from the index, which showed an 80 % sensitivity and 81 % specificity. Area under the curve was 0.889. We offer evidence that, with further refinements, patterns of serum biomarkers might help identify those children more or less likely to have LNB, perhaps ultimately decreasing the need for lumbar punctures.