Clinical features of COVID-19 are diverse, and a useful tool for predicting clinical outcomes based on clinical characteristics of COVID-19 is needed. This study examined the laboratory values and trends that influence mortality in hospitalised COVID-19 patients. Data on hospitalised patients enrolled in a registry study in Japan (COVID-19 Registry Japan) were obtained. Patients with records on basic information, outcomes, and laboratory data on the day of admission (day 1) and day 8 were included. In-hospital mortality was set as the outcome, and associated factors were identified by multivariate analysis using the stepwise method. A total of 8860 hospitalised patients were included. The group with lactate dehydrogenase (LDH) levels >222 IU/L on day 8 had a higher mortality rate compared to the group with LDH levels ≤222 IU/L. Similar results were observed in subgroups formed by age, body mass index (BMI), underlying disease, and mutation type, except for those aged <50 years. When age, sex, BMI, underlying disease, and laboratory values on days 1 and 8 were tested for factors strongly associated with in-hospital mortality, LDH on day 8 was most strongly associated with mortality. LDH level on day 8 was the strongest predictor of in-hospital mortality in hospitalised COVID-19 patients, indicating its potential usefulness in post-treatment decision-making in severe COVID-19 cases.