Objectives The reported number of syphilis cases is increasing in Japan. In this study, we aimed to estimate both the incidence of infection and diagnosis-and-reporting rate using a mathematical model that captures the time course of infection.

Methods We analysed yearly notification data from 1961 to 2016 and stratified according to clinical stage and sex. We built a compartmental ordinary differential equations model to describe the natural epidemiological course of syphilis in which the incidence of infection and diagnosis-and-reporting rate were included as time-varying parameters that we estimated via the maximum likelihood method.

Results From 2008 to 2016, the estimated number of new syphilis cases among men and women was 1070 (95% CI 1037 to 1104) and 302 (95% CI 287 to 318), respectively, which was increased from the previous period (1999–2007) with 269 (95% CI 256 to 282) and 71 (95% CI 64 to 78) cases, respectively. The diagnosis-and-reporting rate did not vary greatly over time. We estimated the number of undiagnosed syphilis cases from either incubating or early clinical (ie, primary or secondary) infections in 2016 to be 262 (95% CI 249 to 275) and 79 (95% CI 74 to 84) for men and women, respectively.

Conclusions The recent increase in reported cases of syphilis is owing to an increase in the number of infections. We found no evidence of time-dependent improvement in detection or reporting.