I doubt that it is very precise. I decided for myself to use only the proportion to compare which terms come up more frequently.
It never seems to be too accurate, but is a good baseline against other keywords. If you make the assumption that it is off by 10-15% across the board, then you can (at least) use it as a weight/measure to see how other terms stack up against it. I believe (correct me if wrong) that the ranking traffic is based on previous calendar month (30-day trend/window).
Always amazes me to read that people think Google would give them bogus numbers. Why would they? But even if you think the numbers are wrong, as Thomas said, you can use them to weigh them against other terms. You also need to understand what the numbers mean, which few seem to. The numbers shown are for the previous month. It does not forecast what will happen, only what has happened. You also need to understand the difference between the different match types.