British aid worker Linda Norgrove may have been accidentally killed by US forces during a rescue mission in Afghanistan. It's not the first time such a perilous task has ended tragically. So how are they planned and what are the key considerations?
Questions were already being asked as to why the operation failed, after it was initially reported that she had been killed when one of her captors detonated a suicide vest.
UK Prime Minister David Cameron's announcement that the 36-year-old Scot may have been killed by a US grenade has now put the mission under closer scrutiny.
All such missions are inherently dangerous. But how do special forces minimise the risks?
International security expert Bob Ayers says the first consideration is which kind of group has the hostage.
In the worst-case scenarios, it could be a group with a reputation for beheading hostages for propaganda purposes.
"If the probability is that the person will be killed then even though it may be very risky, you're still going to give it a try because you've got nothing to lose," he says.
Groups of bandits interested only in a ransom are likely to be less sophisticated and will need to make contact to achieve their aim, says intelligence analyst Glenmore Trenear-Harvey.
He says good, current intelligence is vital and that Nato forces have "far more resources than the bad guys".
Satellites and unmanned drones can accurately locate relevant buildings, while mini-drones flying at 2,000ft can "peer in the doors" at an angle.
Monitoring of mobile phones, radio signals and computer communication, along with information from agents, can also offer an advantage.
"They will build up the profile of the likely captors, their habitat, what weapons are available and any previous modus operandi," he says.
The chiefs of relevant security forces would then be briefed on the planned operations, along with political leaders who would have to sign off any rescue bid.
Mr Ayers, who spent 30 years as an intelligence officer with the US Department of Defense, says factors such as the weather can affect missions - impeding the use of helicopters.
Another consideration is the terrain.
"If you're trying to rescue someone being held inside a castle and you have to climb over walls and run 500 feet across an open area to get there, it will take a long time and typically they will be killed before you get there," he says.
Medics may be needed in case hostages are wounded.
But he adds: "You want to use the minimum number of people. If you increase the numbers of people, you increase the complexity... and the more likely it is that someone will drop the ball along the way."
Mr Trenear-Harvey says this makes co-ordination of all those involved crucial.
"Such operations are by their nature clandestine and there is a chance that the rescuers can come into contact with regular forces who don't know they are there - increasing the chances of friendly fire incidents."
Mr Cameron said that in Linda Norgrove's case, it had been feared she would be "passed up the terrorist chain", increasingly the likelihood she would be killed.
The further from the original kidnappers the hostage is taken, the more difficult the rescue will be, Mr Ayers agrees.
"Typically, you want to do the rescue operation as quickly as you can, with as much speed as possible because you want the captors to be confused as to how to respond."
However, he says this is difficult because every operation is unique.
And no matter how well-planned is the operation, unexpected deviations are almost inevitable.
"You tend to have to make a lot of adjustments. You need people who are well trained at assessing situations and making spot decisions," he adds.