I believe that in the UK students have a master's degree before starting a PhD program (whereas in the US they usually don't), so it isn't quite comparable.
Some do and some don't. It used to be that you needed a 1st class degree in the subject from a good university to be taken straight in. With 2(1) or not a top uni, you would be 'encouraged' to do masters first, which worked as an additional filter.
It seems now you can go directly from a BS into a PhD program - in the past you had to have a master degree first before you were allowed into PhD programs.
Still many programs in the UK require you to have a post-grad / master degree before accepting you into their PhD programs, but you can often start research during your master and expand your master thesis into the PhD (depending on the subject).
In the UK most non-teaching/non-resident PhD programs take 2 years (US equiv. 4years).
The quickest route to an PhD equivalent is a German Doctorate of Law that you can do after your first German Law exam (min study 4years avg. 5years - equiv. to an U.S. JD) and followed by the second German law exam (additional 2years equiv. to an U.S. Bar). Most do their German Doctorate of Law in less than a year - some in 6 months - between the first and the second law exam - hence the large amount of lawyers in Germany with doctoral degrees.
"I believe that in the UK students have a master's degree before starting a PhD program"
Definitely not true - unless this has changed recently. I went pretty much directly from a 4 year BSc(Hons) to working as an Research Associate while attempting a PhD.
From looking around it appears that the research councils now require a Masters degree before they will fund you for a full-time PhD studentship - but that's the funding body, not the entrance requirement for a particular university/department. I'm pretty sure that didn't used to be the case (I considered doing a PhD via a research council studentship before I got offered an RA post).
Note my knowledge is probably wildly out of date - my experience in academia was from '89-'95.
There might also be other sources of funding that don't have this requirement - unless you are converting from another area and need the Masters to get up to speed I'm not sure what it would achieve. All of the people I knew who did have an MSc before doing a PhD had done a "conversion" MSc.
To make up for the amazing improvement in teaching and ever rising school exam results over the last 20years we have had to add an extra year to the traditional 3year undergrad course. For some reason we don't understand these increasingly brilliant school students don't seem to have done any calculus for example before arriving for a physics degree.
The extra 'honours' year of the degree is normally called a MSci/Meng - to differentiate it from a real MSc masters degree. There is usually an option to stop after 3 years and get a normal BSc. Most research councils require the four year version.
The reason British PhDs were nominally 3years is that UK ugrad programs were totally in the nominal subject, there is no minor or required arts/humanities courses - you just do a single subject every lecture.
Similarly a PhD traditionally has no taught courses or required extra material, just do the research 24x7. And since the research councils gave you a grant you could live on there was no need to teach/TA more than a nominal amount for experience (and beer money)
