One thing that I really annoys me is the common misconception that Tamiflu should only be used within 48 hours of symptom onset and is only able to diminish symptoms. While the strongest data supports the above two claims, there is evidence that Tamiflu confers additional--and potentially lifesaving--benefits.
What must be kept in mind is that when Tamiflu received FDA approval it was studied in uncomplicated influenza cases and was shown to be maximally (not exclusively) beneficial when administered within 48 hours of symptom onset. Severe cases were not included in those randomized controlled trials and, consequently, strong data regarding pneumonia and other severe complications of influenza do not really exist.
The other element of this story is that because severe influenza is unequivocally life-threatenening it is not possible to conduct a placebo-controlled trial as withholding antiviral treatment from a severely ill individual can not be justified.
While strong data (i.e. prospective, randomized, placebo-controlled) may not exist that does not mean no data exist. Retrospective data has shown Tamiflu to be beneficial in severe cases, especially when given early.
M y practice is to use Tamiflu (or Relenza) for influenza cases irrespective of how long symptoms have occurred, especially in those at high risk of a severe course.