In 1996 the American College of Sports Medicine (ACSM) released guidelines that athletes "should drink as much as tolerable" during physical activity and this advice has been repeated to athletes over the world, to stay ahead of the thirst curve - but is this essential for optimum health and performance?
Drinking guidelines for exercise: what evidence is there that athletes should drink "as much as tolerable", "to replace the weight lost during exercise" or "ad libitum"?
Drinking policies and exercise-associated hyponatraemia: is anyone still promoting overdrinking?
Is drinking to thirst optimum?
The above reviews suggest that the ACSM guidelines and others who promote similar advice are not evidence based and adoption of these standards may lead to weight gain, water retention and in severe cases death from exercise-associated hyponatraemic encephalopathy.
The last review also compares the strategy of drinking to avoid weight loss (0% dehydration) to drinking ad libitum (drinking when you are thirsty) and concludes "Drinking ad libitum appears to optimize performance and safety during exercise in many situations. The presence of thirst, not of water loss, may be the biological signal that impairs exercise performance in those who drink less than their thirst dictates during exercise."
To answer the original question, drink when you are thirsty. There is no evidence to suggest that drinking as much as possible or drinking to avoid weight loss has any performance or health benefits (in fact they may even have the potential to cause more harm than good).