Theileria parva is a protozoan parasite transmitted by the brown ear tick Rhipicephalus appendiculatus that causes East Coast fever (ECF) in cattle, resulting in substantial economic losses in the regions of southern, eastern and central Africa. The schizont form of the parasite transforms the bovine host lymphocytes into actively proliferating cancer-like cells. However, how T. parva causes bovine host cells to proliferate and maintain a cancerous phenotype following infection is still poorly understood. On the other hand, current efforts to develop improved vaccines have identified only a few candidate antigens. In the present paper, we report the first comparative transcriptomic analysis throughout the course of T. parva infection. We observed that the development of sporoblast into sporozoite and then the establishment in the host cells as schizont is accompanied by a drastic increase of upregulated genes in the schizont stage of the parasite. In contrast, the ten highest gene expression values occurred in the arthropod vector stages. A comparative analysis showed that 2845 genes were upregulated in both sporozoite and schizont stages compared to the sporoblast. In addition, 647 were upregulated only in the sporozoite whereas 310 were only upregulated in the schizont. We detected low p67 expression in the schizont stage, an unexpected finding considering that p67 has been reported as a sporozoite stage-specific gene. In contrast, we found that transcription of p67 was 20 times higher in the sporoblast than in the sporozoite. Using the expression profiles of recently identified candidate vaccine antigens as a benchmark for selection for novel potential vaccine candidates, we identified three genes with expression similar to p67 and several other genes similar to Tp1-Tp10 schizont vaccine antigens. We propose that the antigenicity or chemotherapeutic potential of this panel of new candidate antigens be further investigated. Structural comparisons of the transcripts generated here with the existing gene models for the respective loci revealed indels. Our findings can be used to improve the structural annotation of the T. parva genome, and the identification of alternatively spliced transcripts.