A Cochrane review examining the effects of commonly used oral analgesics found that no drug produced reliable pain relief in every patient.
Researchers hope the findings will help doctors and patients decide which drugs give the greatest likelihood of effective pain relief.
The meta-analysis of 35 previous Cochrane reviews brought together data on 45,000 patients from 350 studies. Dr Andrew Moore and colleagues at Oxford University examined how well many different analgesics work against postoperative pain.
They found effectiveness varied greatly. More than 70% of patients had good pain relief with drugs such as 500mg paracetamol plus 200mg ibuprofen and 120mg etoricoxib.
But just 35% benefited from 1000mg aspirin and 600mg paracetamol alone. The worst performing drug was codeine, with just 14% of patients achieving good pain relief. The length of pain relief also varied from two to 20 hours.
Dr Moore said: 'Pain relief doesn't have to be a mystery. There is a body of reliable evidence about how well 46 different drug/dose combinations work against acute pain.'
Dr Bill Beeby, GPC prescribing subcommittee chairman, said the review could be 'very useful' for GPs. 'It sounds like it will complement with science the existing pain ladder,' he said.