Ammopiptanthus mongolicus (Maxim. Ex Kom.) Cheng f., a relic tree of the Tertiary period, plays a critical role in maintaining desert ecosystems in the Mid-Asia region. Genome-scale gene expression profiling studies will provide deep insight into the molecular mechanism underlying the drought tolerance of A. mongolicus. In the present study, we investigated the transcriptional changes induced by drought treatment in A. mongolicus leaves by establishing a comprehensive transcriptome database and then performing a Digital Gene Expression (DGE) analysis using Solexa sequencing technology. A comprehensive transcriptome database was obtained by assembling the Illumina unigenes with expressed sequence tags (EST) available publicly, and other high throughput sequencing data. To analyze the dynamic and complicated gene regulation network during PEG6000-induced drought treatment in leaves of A. mongolicus, a time-course gene expression analysis was performed using tag-based DGE technology, which identified 437, 1,247 and 802 differentially expressed transcripts in 1, 24 and 72 h drought stress libraries, respectively. GO and KEGG analyses revealed hormone signal transduction and phenylpropanoid biosynthesis were enriched during drought treatment. A batch of drought-regulated transcription factor transcripts were identified, including the subsets of HD-ZIP, bZIP, WRKY, AP2/ERF and bHLH family members, which may play roles in drought response in A. mongolicus. The sequence collection assembled in the present study represents one of the most comprehensive transcriptome databases for A. mongolicus currently. The differentially expressed transcripts identified in our study provide a good start for identifying the key genes in stress response and performing functional analysis to reveal their roles in stress adaptation in planta.