Abstract Background Television viewing time is associated cross-sectionally with abnormal glucose tolerance and diabetes risk; however, the impact of changes in television viewing time on glycaemic measures is less understood. We examined relationships of 5-year change in television viewing time with 5-year change in glucose homeostasis markers. Methods Participants in the Australian Diabetes, Obesity and Lifestyle study with data available at the 1999-2000 baseline and the 2004-2005 follow-up were included (4870; 45% men). Television viewing time (h/week) was assessed by questionnaire. Fasting plasma glucose, serum insulin and 2-h plasma glucose were obtained from an oral glucose tolerance test. Beta-cell function and insulin resistance were ascertained using the homeostasis model assessment 2-calculator. Associations of change in television viewing time with changes in glucose homeostasis markers were examined using linear regression models [beta-coefficients (95% CI)]. Adjustments included baseline measures of age, television viewing time and glycaemic marker, and baseline and 5-year change in diet quality, energy intake, physical activity and waist circumference. Results For every 5-h per week increase in television viewing time from baseline to 5-year follow-up, changes in glucose homeostasis markers were observed: among women there was a significant increase in fasting plasma glucose [0.01 (0.000.02) mmol/l] insulin resistance [0.03 (0.010.05)] and insulin secretion [1.07 (0.022.12) %]; insulin levels increased [men: 1.20 (0.302.09); women: 1.06 (0.321.80) pmol/l]; in men, 2-h plasma glucose levels increased [0.06 (0.010.1) mmol/l]. Conclusion Five-year increases in television viewing time were associated adversely with changes in glucose homeostasis markers. These findings add to earlier cross-sectional evidence that television viewing time can be associated with biomarkers of diabetes risk.