Noise pollution is reported to be associated with diabetes, but few studies have elucidated the associations between noise frequency characteristics. We aimed to evaluate the relationships between different noise frequency components and incident hyperglycaemia. An industry-based cohort of 905 volunteers was enrolled and followed up to 2012. Octave-band frequencies of workstation noise and individual noise levels were measured in 2012 to classify subjects' exposures retrospectively. We applied Cox regression models to estimate the relative risk (RR) of hyperglycaemia. An increased RR for hyperglycaemia of 1.80 (95% confidence interval [CI]: 1.04, 3.10) was found among subjects exposed to ≥ 85 A-weighted decibels (dBA) compared with those exposed to < 70 dBA. The high-exposure groups at frequencies of 31.5, 63, 125, 250, 500, 1000, and 2000 Hz had a significantly higher risk of hyperglycaemia (all p values < 0.050) than the low-exposure groups. A 5-dB increase in noise frequencies at 31.5, 63, 125, 250, 500 Hz, and 1000 Hz was associated with an elevated risk of hyperglycaemia (all p values < 0.050), with the highest value of 1.27 (95% CI: 1.10, 1.47) at 31.5 Hz (p = 0.001). Occupational noise exposure may be associated with an increased incidence of hyperglycaemia, with the highest risk observed at 31.5 Hz.