Lakes play a significant role in the global carbon cycle, acting as sources and sinks of carbon dioxide (CO2). In situ measurements of CO2 flux (FCO2) from lakes have generally been collected during daylight, despite indications of significant diel variability. This introduces bias when scaling up to whole-lake annual aquatic carbon budgets. We conducted an international sampling program to ascertain the extent of diel variation in FCO2 across lakes. We sampled 21 lakes over 41 campaigns and measured FCO2 at 4-h intervals over a full diel cycle. Rates of FCO2 ranged from -3.16 to 4.39 mmol m-2 h(-1). Integrated over a day, FCO2 ranged from -381.68 to 878.49 mg C m(-2) d(-1) (mean = 76.54) across campaigns. We identified three characteristic diel patterns in FCO2 related to trophic status and show that for half of the campaigns, daily flux estimates were biased by > 50% if based on a single (daytime) measurement.