Objective There have been no previous studies regarding the relationship between the respiratory function and incident diabetes in East Asian women in whom obesity is not prevalent. Methods Using Cox regression models, hazard ratios (HRs) for incident diabetes for each one SD increase in the percent vital capacity (%VC), a parameter of restrictive pulmonary dysfunction, forced expiratory volume in one second over forced vital capacity (FEV1/FVC), a parameter of obstructive pulmonary dysfunction, and log high-sensitivity C-reactive protein (hs-CRP) were calculated in 1,704 men and 1,016 women among a Japanese health screening population. Results Over the four-year study period (mean of 3.4 years), 54 men (3.2%) and 19 women (1.9%) developed diabetes. The HRs [95% confidence intervals (CIs); p values] for incident diabetes for each one SD increase in %VC and log hs-CRP were 0.81 (0.61-1.06; 0.127) and 1.29 (0.97-1.73; 0.085), respectively in men, and 0.55 (0.35-0.88) 0.012 and 2.50 (1.37-4.57; 0.003), respectively in women adjusted for age, fasting glucose and body mass index. The FEV1/FVC was not found to be significantly associated with incident diabetes in any regression models. The Spearman's correlation coefficients between %VC and hs-CRP was -0.11 (p<0.001) in men and -0.08 (p=0.007) in women. The HRs (95% CIs; p values) for each one SD increase in %VC and hs-CRP were 0.86 (0.65-1.13; 0.278) and 1.26 (0.94-1.70; 0.125), respectively in men, and 0.56 (0.34-0.94; 0.028) and 2.95 (1.49-5.81; 0.002), respectively in women, further adjusted for each other, systolic blood pressure, triglycerides and HDL cholesterol. Conclusion A low %VC was found to be associated with incident diabetes, independent of hs-CRP in women among a Japanese health screening population.