When you use a USB port on a notebook computer to charge, for example, an iPod or a Blackberry, how efficient is the transfer of power from the notebook’s battery to the device (assuming that the notebook would be on anyway)?
Does the efficiency vary if the notebook is plugged in to the mains?
Charging when connected to the mains is more efficient, because charging the notebook battery incurs charging losses, and self discharge if there is a long delay between charging and use.
Notebook batteries are made for high capacity and light weight, not efficiency. Lead-acid batteries have the highest charge/discharge efficiency at around 85%, which is why (well, price too) they still remain the first choice for solar storage, even though they are almost the oldest of the still used battery technologies. Not sure how LiIon batteries compare, but I would swag 60-75%.
The power supply in the notebook probably supplies the USB ports at 90% efficiency or so, under load. a lot of that will be fixed losses, that you incur even at no load though, so if you are operating the notebook anyway, you probably see around 95% efficiency to the computer port.
How efficient the charging circuit in your portable device is antibody’s guess however. A simple linear regulator could well be worse than 50%.
My immune system has no idea, either. 
Good answer, however, but I can’t help wondering if the OP isn’t using “efficiency” in the more colloquial sense of “how fast will it charge?” That is going to depend on whether you have USB 1.1 or USB 2.0 ports and the terminal voltage, type and capacity of your device’s battery, as well as current limiting inherent in the device’s charging circuitry, among other factors. USB 1.1 specifies a max supply current of 100 mA at 5 VDC, while USB 2.0 specifies up to 500 mA (though some ports may supply less, depending upon the manufacturer), also at 5 VDC. Typically, portable device batteries are trickle-charged at a few to several tens of milliamps over a few to several hours, depending on capacity, so almost any port should provide plenty of current; the exact charging time is going to depend heavily on several factors, however.
Thanks Kevbo, that’s a good starting point.
No, I’m using efficiency in the sense of what proportion of the energy stored within the notebook battery is transferred to the device.
I’m considering policies for reducing energy usage in my office. The big effect of getting people to charge devices (more) directly from the mains will be to make them less likely to leave their computers on, but it’d be good to put a figure on the transfer efficiency for persuasion purposes.