The dinoflagellate Akashiwo sanguinea is a harmful algal species and commonly observed in estuarine and coastal waters around the world. Harmful algal blooms (HABs) caused by this species lead to serious environmental impacts in the coastal waters of China since 1998 followed by huge economic losses. However, the full-length transcriptome information of A. sanguinea is still not fully explored, which hampers basic genetic and functional studies. Herein, single-molecule real-time (SMRT) sequencing technology was performed to characterize the full-length transcript in A. sanguinea. Totally, 83.03 Gb SMRT sequencing clean reads were generated, 983,960 circular consensus sequences (CCS) with average lengths of 3,061 bp were obtained, and 81.71% (804,016) of CCS were full-length non-chimeric reads (FLNC). Furthermore, 26,461 contigs were obtained after being corrected with Illumina library sequencing, with 20,037 (75.72%) successfully annotated in the five public databases. A total of 13,441 long non-coding RNA (lncRNA) transcripts, 3,137 alternative splicing (AS) events, 514 putative transcription factors (TFs) members from 23 TF families, and 4,397 simple sequence repeats (SSRs) were predicted, respectively. Our findings provided a sizable insights into gene sequence characteristics of A. sanguinea, which can be used as a reference sequence resource for A. sanguinea draft genome annotation, and will contribute to further molecular biology research on this harmful bloom algae.