Glioblastoma (GBM) is a complex disease with high intratumoral heterogeneity, understanding the molecular characteristics of intratumoral heterogeneity accurately is the basis for precision treatment. Although the existing typing strategy based on tumor molecular characteristics has a positive effect, there is still room for improvement, which is mainly because the traditional typing is completed based on the sequencing data of tissue samples, that is, the obtained data are the average level of patient tumor tissues, masking the intratumoral heterogeneity of a single patient and cannot reflect the real level of patient tumor cells. At present, cancer molecular typing is mostly performed based on transcriptome (RNA-seq) without considering lncRNA molecules that are also tissue-specific and developmental stage-specific. Therefore, in this study, we used lncRNAs as typing markers and combined single-cell expression profiles to retype glioblastoma, providing new ideas for GBM molecular typing, and further analyzed the shortcomings of traditional therapies at the singlecell level based on typing results and proposed new precise therapeutic insights. We downloaded GBM single-cell sequencing data from GSE84465 and performed a series of preprocessing. The intratumoral heterogeneity of patients at the single-cell level was revealed using t-SNE, and the room for improvement of the existing traditional histotyping method was revealed using heat map and density curve. Subsequently, to validate the feasibility of lncRNA typing, we compared the similarities and differences of expression patterns between lncRNAs and mRNAs in GBM cells. Then, we used the R package “Seurat” to perform unsupervised clustering of GBM cells for re-typing and performed a detailed analysis of the biological characteristics of each subtype, including differentially expressed lncRNAs and marker lncRNAs. For validation, we performed survival analysis on GBM tissue data from the TCGA database to reveal the impact of different subtypes on patient survival prognosis. Eventually, based on the results, we screened the therapeutic drugs of each subtype by targeting the downstream regulatory genes of lncRNAs and proposed a new precision therapeutic strategy. GBM has significant intratumoral heterogeneity at the single-cell level, with more than one traditional subtype highly expressed in each patient, which reflects the shortcomings of traditional histotyping. LncRNAs and mRNAs have similar expression patterns in GBM cells, and the expression coefficient of variation of lncRNAs is higher than that of mRNAs, meaning that lncRNAs will better reflect the intratumoral heterogeneity. GBM was reclassified into four subtypes by unsupervised clustering, with different subtypes having different biological characteristics. Survival analysis showed that patients with different subtype compositions had different prognostic outcomes, so different subtypes had different effects on patient prognosis. Based on this, 47 drugs were screened for treatment. There are both shared and unique drugs between different subtypes. A new precision treatment strategy was proposed: for patients with different subtypes, in addition to the combination of drugs targeting single subtype, drugs targeting multiple subtypes can also be selected. Intratumoral heterogeneity may lead to poor prognosis or recurrence after treatment, and more precise typing of GBM can be performed based on single-cell lncRNA expression profiles. The biological characteristics possessed by different subtypes will have different effects on patients, such as survival time. For different subtypes, there are both drugs targeting single subtype and drugs targeting multiple subtypes, and we prefer drugs targeting multiple subtypes because this strategy can maximize medication efficiency and reduce the types of medication to reduce risks and side effects.