Within‐host model specified by viral dynamic parameters is a mainstream tool to understand SARS‐CoV‐2 replication cycle in infected patients. The parameter uncertainty further affects the output of the model, such as the efficacy of potential antiviral drugs. However, gathering empirical data on these parameters is challenging. Here, we aim to conduct a systematic review of viral dynamic parameters used in within‐host models by calibrating the model to the viral load data measured from upper respiratory specimens. We searched the PubMed, Embase and Web of Science databases (between 1 December 2019 and 10 February 2022) for within‐host modelling studies. We identified seven independent within‐host models from the above nine studies, including Type I interferon, innate response, humoral immune response or cell‐mediated immune response. From these models, we extracted and analyse seven widely used viral dynamic parameters including the viral load at the point of infection or symptom onset, the rate of viral particles infecting susceptible cells, the rate of infected cells releasing virus, the rate of virus particles cleared, the rate of infected cells cleared and the rate of cells in the eclipse phase can become productively infected. We identified seven independent within‐host models from nine eligible studies. The viral load at symptom onset is 4.78 (95% CI:2.93, 6.62) log(copies/ml), and the viral load at the point of infection is −1.00 (95% CI:−1.94, −0.05) log(copies/ml). The rate of viral particles infecting susceptible cells and the rate of infected cells cleared have the pooled estimates as −6.96 (95% CI:−7.66, −6.25) log([copies/ml]–1 day–1) and 0.92 (95% CI:−0.09, 1.93) day–1, respectively. We found that the rate of infected cells cleared was associated with the reported model in the meta‐analysis by including the model type as a categorical variable (p < .01). Joint viral dynamic parameters estimates when parameterizing within‐host models have been published for SARS‐CoV‐2. The reviewed viral dynamic parameters can be used in the same within‐host model to understand SARS‐CoV‐2 replication cycle in infected patients and assess the impact of pharmaceutical interventions.