Long diagnostic delays (LDDs) in patients with coronavirus disease 2019 (COVID-19) might decrease the effectiveness of patient isolation in reducing subsequent transmission. We assumed that direction of government considerably increased probability of LDD among COVID-19 cases with unknown exposure in Japan. This study aimed to investigate association of route of case detection and proportion of LDD of COVID-19 in Japan. We included confirmed COVID-19 patients with symptom onset between the ninth and eleventh week in 2020, in 6 prefectures of Japan. LDD was defined as the duration between COVID-19 symptom onset and confirmation ≥6 days. We used multivariable logistic regression analyses to elucidate factors associated with LDD. The mean diagnostic delay for 364 cases was 6.3 days. Proportion of LDD was 38% for cases with known exposure, and 65% for cases with unknown exposure. The probability of LDD in cases with unknown exposure was significantly higher than that for known exposure cases (adjusted odds ratio: 2.38, 95% confidence interval: 1.354–4.21). Early PCR test after symptom onset, strengthening of PCR test capacity, and investigations to study impact of high proportion of LDD in cases without known exposure might be necessary.