Abstract Background In infectious disease surveillance, when the laboratory confirmation of the cases is time-consuming, there is often a time lag between the number of suspect cases and the number of confirmed cases. This study proposes a dynamic statistical model to estimate the daily number of new cases and the daily cumulative number of infected cases, which was then applied to historic dengue fever data. Methods The duration between the date of disease onset and date of laboratory confirmation was assumed to follow a gamma distribution or a nonparametric distribution. A conditional probability of a case being a real case among the unconfirmed cases on a given date was then calculated. This probability along with the observed confirmed cases was integrated to estimate the daily number of new cases and the cumulative number of infected cases. Results The distribution of the onset-to-confirmation time for the positive cases was different from that of the negative cases. The daily new cases and cumulative epidemic curves estimated by the proposed method have a lower absolute relative bias than the values estimated solely based on the available daily-confirmed cases. Conclusion The proposed method provides a more accurate real-time estimation of the daily new cases and daily cumulative number of infected cases. The model makes use of the most recent "moving window" of information relative to suspect cases and dynamically updates the parameters. The proposed method will be useful for the real-time evaluation of a disease outbreak when case classification requires a time-consuming laboratory process to identify a confirmed case.