BACKGROUND: A low-carbohydrate diet (LCD) may be effective for reducing cardiovascular disease (CVD) risk factors. However, its association with mortality remains unclear, especially in populations with higher carbohydrate and lower fat intake. OBJECTIVES: This study aimed to evaluate the association between LCD scores and mortality risk in participants from the Japan collaborative cohort study. METHODS: The analysis included 22,659 males and 35,192 females aged 40-79 y at baseline. A food frequency questionnaire was used to construct the LCD scores, which were further categorized into 3 types based on the fat intake source: animal, fish, and plant. The association between the LCD scores and total and cause-specific mortality was estimated using the Cox proportional hazards model adjusted for confounders. RESULTS: During a median follow-up period of 19.4 y, 10,835 deaths (5835 males and 5000 females) were recorded. For all participants, the overall LCD score was inversely associated with total mortality. The lowest hazard ratio (95% confidence interval) was 0.85 (0.78, 0.93) in the eighth decile (P-linear trend = 0.03). Fish fat-based LCD score had a U-shaped association with total mortality (P-nonlinear trend = 0.01). An inverse association was observed with CVD mortality (P-linear trend = 0.04 of overall LCD score
P-linear trend = 0.04 of fish fat-based LCD score). In females, there was an inverse association between the overall LCD score and total mortality. A U-shaped association was also shown for fish fat-based LCD score. The fish fat-based LCD score was inversely associated with CVD mortality. CONCLUSIONS: Moderate overall and fish fat-based LCD scores were associated with a lower risk of total mortality in the Japanese population. Fish fat-based LCD score had a significant inverse association with CVD mortality.