There is also a discussion in the context of social networks and the successful patterns.
Startups generally has been very successful in markets where the solution they offer can -
Facebook is the latest incarnation of services to "connect everyone together" efficiently (and make the process fun).
Companies that have successfully done so had grown very fast in the past - e.g. Skype, Nokia (Connecting people was their slogan).
The general observation of why such a value exist is Metcalfe's Law, which states that "the value of a telecommunications network is proportional to the square of the number of connected users of the system (n2) "
While back in telecommunications world - the "user generated content" was conversation, in the current paradigm, its status updates, photos, tags, links, textual conversation, twitter updates, game achievements, etc.
If such a pace of sharing continues, you can sure to expect that more aspects would be shareable in the near future (and probably wonder about what cannot be shared).
If your question is about "technology" sector in general, as opposed to specifically software startups, then without a doubt Biotech has been the most successful. Biotech and health care are still a growing market and the economy, while it's had some impact, has had less of an impact there than in other areas due to the real demand and government actions.
If you're asking specifically about software, there is evidence to show that software in healthcare is also more successful than average. The same is true of education software. See some data here from Professor Scott Shane: http://boss.blogs.nytimes.com/2009/10/12/which-industries-are-most-favorable-to-start-ups/ about startup failure rates. It's the inverse of your question, in a way, but it indicates what sectors are more likely to have startups that don't fail.
Scott Shane of Case Western Reserve University here in Ohio is an internationally recognized researcher on entrepreneurship. I recommend you check out his work if you're interested in hard data on this. http://wsomfaculty.cwru.edu/shane/ Please note that he does not focus on tech or software specifically, though, but rather on new business ventures in general. This may or not be of use in your individual case.
Best of luck!
This question also depends on your geographic location.
In the far east and south Asia you are seeing Mobility startups doing really well these days as mobile adoption is growing at a rapid pace.
In the US social media is definitely a space which is very hot but I don't see as many companies being able to monetize their services effectively.
One of the hotter spaces is definitely iPhone and mobile apps leveraging off their app stores.
The holistic trend line seems to be leveraging off existing platforms which have an exponentially growing base.
Google "augmented reality" and/or "web squared" :)
Rather than try to jump on a bandwagon or hit a moving target you are better off choosing something you are interested in or passionate about. Just trying to go where the money is makes you a follower - and that is not a good start for starting a company in my opinion.
There really is no delineation - people see needs or have ideas of things they want for themselves and then develop for it.
Find a "pain point" and fill it.
That depends on how you want to define Successful, and what timeframe you are interested in.
For example, Twitter may be considered successful, but it is doubtful to me that they will ever be profitable.
Google had an advantage of starting when there was more money available for startups, during the tech bubble, so their challenges were substantially different than now, IMO.
I think the "trend" in successful startups is that they are smart enough to understand the markets they're in, identify where that market is going, size up the opportunity, determine how to monetize the opportunity with a product or service that delivers value, and get there before the competition.
To answer your general question:
Most successful start-ups have failed atTwitter with Odeo for example. Or in Google's case they release early and interate until it's correct.