The aim was to examine the effects of replacing self‐reported leisure‐time sedentary behavior with sleep, light‐to‐moderate physical activity, or vigorous physical activity on incident diabetes among Danish adults using isotemporal substitution modeling. Participants ≥25 years from the Danish Capital Region Health Survey 2007 (N = 69 800, response rate 52.3%), 2010 (N = 95 150, response rate 52.3%), and 2013 (N = 95 150, response rate 43.5%) were included. Information on daily sleep duration, leisure‐time sedentary behavior, and movement behaviors was collected by questionnaire. Information on incident diabetes was obtained from National registers. Analyses included Cox proportional hazards regression models and isotemporal substitution analyses, with time (in years) from baseline to incident diabetes or censoring December 31, 2017. Potential confounders, sex, age, BMI, ethnicity, education, smoking, inflammatory joint disease, perceived stress, physical and mental component scale and work status, were included. Out of N = 87 339 in the final study sample, n = 3007 had incident diabetes during a mean follow‐up time of 7.4 years. Adults with incident diabetes included more men, higher mean age, and higher BMI, compared to respondents without incident diabetes. Theoretically substituting 30 minutes of leisure‐time sedentary behavior with light‐to‐moderate PA (HR: 0.96; 95% CI: 0.94; 0.98) or with vigorous PA (HR: 0.82; 95% CI: 0.72; 0.94) decreased the risk of incident diabetes. We found no change in incident diabetes risk of substituting sedentary time with sleep (HR: 1.00; 95% CI: 0.97; 1.02). Substituting 30 minutes per day of leisure‐time sedentary behavior with light‐to‐moderate or vigorous PA may significantly reduce the risk of incident diabetes among adults.