Analysis of gene transcription patterns in complex tissues with multiple cell types is a major challenge. Examination of cellular subpopulations for molecular expression patterns requires their isolation from other surrounding cells. We performed single-cell mRNA analysis to study gangliogliomas obtained from patients with pharmacoresistant epilepsy (n ¼ 6), in order to characterize CD34 expressing cells found in these tumors. Fresh-frozen biopsy tissue was analyzed by initial in situ-reverse transcription (in situ-RT) with oligonucleotides, subsequent immunohistochemistry (IHC) to identify specific cell types, and laser-capture microdissection (LCM, herein termed immuno-LCM) to obtain antigen-expressing cell subpopulations. Isolated complementary DNAs (cDNAs) were then quantified by real time-polymerase chain reaction (RT-PCR). We found that short-vs long-term incubation time for the IHC step did not adversely affect cDNA abundance obtained by subsequent RT-PCR, either for high-abundance (glyceraldehyde dehydrogenase; GAPDH), medium-abundance (glial fibrillary acidic protein; GFAP), or low abundance (neurofilament; NFM) gene transcripts. We also determined that the cellular specificity of capture was excellent, as determined by lack of contamination between different immuno-LCM cell isolates. We were therefore able to examine the lineage expression markers of isolated CD34-expressing cells. We observed coexpression of CD34 and NFM, suggesting neuronal differentiation of the CD34 expressing cellular elements in gangliogliomas. Expression markers for other cellular types (myelin basic protein for oligodendroglia; GFAP for astrocytes) were negative. Our findings support the hypothesis that gangiogliomas contain neuronal elements with compromised or atypical differentiation. We consider that this in situ-RT/immuno-LCM protocol is of general applicability, whereby virtually any primary antibody can be used to facilitate capture of individual cells in tissue sections for molecular analysis.