Coronavirus disease (COVID-19) is a heterogeneous infectious disease caused by the novel coronavirus, severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). The distinctive features of COVID-19 infection encompass a clinical spectrum ranging from asymptomatic forms to severe illness involving diverse underlying pathophysiological processes such as hyperinflammation, endothelial dysfunction, thrombotic microangiopathies, and multiple organ failure. However, the host responses to COVID-19, including systemic inflammation and immunity against SARS-CoV-2, remain largely undefined across different disease states. Here, we performed global transcriptome profiling of 228 peripheral blood samples of COVID-19 patients with different clinical manifestations and subsequent clustering using an unsupervised algorithm. We identified various transcriptomic clusters and globally dysregulated immune-related pathways, leading to distinct host inflammatory cytokine profiles in severe and critical SARS-CoV-2 infected patients. This underscores the association between COVID-19 pathogenesis and excessive cytokine release. Additionally, our results reveal variances in circulating cell populations and the activation of apoptosis and the p53 signaling pathway in lymphocytes induced by SARS-CoV-2 infection. The transcriptome dataset of COVID-19 patients with diverse clinical presentations offers valuable insights into molecular mechanisms and may inform clinical decisions on anti-inflammatory therapies.